LCO No. 1489 1 of 37
General Assembly Committee Bill No. 2
February Session, 2024
LCO No. 1489
Referred to Committee on GENERAL LAW
Introduced by:
(GL)
AN ACT CONCERNING ARTIFICIAL INTELLIGENCE.
Be it enacted by the Senate and House of Representatives in General
Assembly convened:
Section 1. (NEW) (Effective October 1, 2024) For the purposes of this 1
section and sections 2 to 7, inclusive, of this act, unless the context 2
otherwise requires: 3
(1) "Algorithmic discrimination" means any condition in which an 4
automated decision tool materially increases the risk of any unjustified 5
differential treatment or impact that disfavors any individual or group 6
of individuals on the basis of their actual or perceived age, color, 7
disability, ethnicity, genetic information, limited proficiency in the 8
English language, national origin, race, religion, reproductive health, 9
sex, veteran status or other classification protected under the laws of this 10
state; 11
(2) "Artificial intelligence" means any technology, including, but not 12
limited to, machine learning, that uses data to train an algorithm or 13
predictive model for the purpose of enabling a computer system or 14
Committee Bill No. 2
LCO No. 1489 2 of 37
service to autonomously perform any task, including, but not limited to, 15
visual perception, language processing or speech recognition, that is 16
normally associated with human intelligence or perception; 17
(3) "Artificial intelligence system" means any machine-based system 18
that, for any explicit or implicit objective, infers from the inputs such 19
system receives how to generate outputs, including, but not limited to, 20
content, decisions, predictions or recommendations, that can influence 21
physical or virtual environments; 22
(4) "Automated decision tool" means any service or system that (A) 23
uses artificial intelligence, and (B) has been specifically developed and 24
marketed, or specifically modified, to make, or be a controlling factor in 25
making, any consequential decision; 26
(5) "Consequential decision" means any decision that has a material 27
legal or similarly significant effect on any consumer's access to, or the 28
availability, cost or terms of, any criminal justice, education enrollment 29
or opportunity, employment or employment opportunity, essential 30
good or service, financial or lending service, government service, health 31
care service, housing, insurance or legal service; 32
(6) "Consumer" means any individual who is a resident of this state; 33
(7) "Deploy" means to use a generative artificial intelligence system 34
or high-risk artificial intelligence system; 35
(8) "Deployer" means any person doing business in this state that 36
deploys (A) a generative artificial intelligence system, or (B) a high-risk 37
artificial intelligence system; 38
(9) "Developer" means any person doing business in this state that 39
develops, or intentionally and substantially modifies, (A) a generative 40
artificial intelligence system, or (B) a high-risk artificial intelligence 41
system; 42
(10) "General purpose artificial intelligence model" (A) means any 43
Committee Bill No. 2
LCO No. 1489 3 of 37
form of artificial intelligence system that (i) displays significant 44
generality, (ii) is capable of competently performing a wide range of 45
distinct tasks, and (iii) can be integrated into a variety of downstream 46
applications or systems, and (B) does not include any artificial 47
intelligence model that is used for development, prototyping and 48
research activities before such model is released on the market; 49
(11) "Generative artificial intelligence system" means any artificial 50
intelligence system, including, but not limited to, a general purpose 51
artificial intelligence model, that is able to produce synthetic digital 52
content; 53
(12) "High-risk artificial intelligence system" means any artificial 54
intelligence system that, when deployed, makes, or is a controlling 55
factor in making, a consequential decision; 56
(13) "Intentional and substantial modification" means any deliberate 57
change made to (A) a generative artificial intelligence system, other than 58
a change made to a generative artificial intelligence system as a result of 59
learning after the generative artificial intelligence system has been 60
deployed, that (i) affects compliance of the generative artificial 61
intelligence system, or (ii) changes the purpose of the generative 62
artificial intelligence system, or (B) a high-risk artificial intelligence 63
system that creates, or potentially creates, any new risk of algorithmic 64
discrimination; 65
(14) "Machine learning" means any technique that enables a computer 66
system or service to autonomously learn and adapt by using algorithms 67
and statistical models to autonomously analyze and draw inferences 68
from patterns in data; 69
(15) "Person" means any individual, association, corporation, limited 70
liability company, partnership, trust or other legal entity; 71
(16) "Synthetic digital content" means any digital content, including, 72
but not limited to, any audio, image, text or video, that is produced by 73
Committee Bill No. 2
LCO No. 1489 4 of 37
a generative artificial intelligence system; and 74
(17) "Trade secret" has the same meaning as provided in section 35-75
51 of the general statutes. 76
Sec. 2. (NEW) (Effective October 1, 2024) (a) Beginning on July 1, 2025, 77
each developer of a high-risk artificial intelligence system shall use 78
reasonable care to protect consumers from any known or reasonably 79
foreseeable risks of algorithmic discrimination. In any enforcement 80
action brought on or after said date by the Attorney General pursuant 81
to section 7 of this act, there shall be a rebuttable presumption that a 82
developer used reasonable care as required under this subsection if the 83
developer complied with the provisions of this section. 84
(b) Beginning on July 1, 2025, and except as provided in subsection 85
(f) of this section, no developer of a high-risk artificial intelligence 86
system shall offer, sell, lease, license, give or otherwise provide to a 87
deployer a high-risk artificial intelligence system unless the developer 88
makes available to the deployer: 89
(1) A general statement describing the intended uses of such high-90
risk artificial intelligence system; and 91
(2) Documentation (A) disclosing (i) all known or reasonably 92
foreseeable limitations of such high-risk artificial intelligence system, 93
including, but not limited to, any known or reasonably foreseeable risks 94
of algorithmic discrimination arising from the intended uses of such 95
high-risk artificial intelligence system, (ii) the purpose of such high-risk 96
artificial intelligence system and the intended benefits, uses and 97
deployment contexts of such high-risk artificial intelligence system, and 98
(iii) a summary of the type of data, if any, intended to be collected from 99
consumers and processed by such high-risk artificial intelligence system 100
when such high-risk artificial intelligence system is deployed, and (B) 101
describing (i) the type of data used to program or train such high-risk 102
artificial intelligence system, (ii) how such high-risk artificial 103
intelligence system was evaluated for performance and relevant 104
Committee Bill No. 2
LCO No. 1489 5 of 37
information related to explainability before such high-risk artificial 105
intelligence system was offered, sold, leased, licensed, given or 106
otherwise provided to a deployer, (iii) the data governance measures 107
used to cover the training datasets and the measures used to examine 108
the suitability of data sources, possible biases and appropriate 109
mitigation, (iv) the intended outputs of such high-risk artificial 110
intelligence system, (v) the measures the developer has taken to mitigate 111
any known or reasonably foreseeable risks of algorithmic discrimination 112
that may arise from deployment of such high-risk artificial intelligence 113
system, and (vi) how a consumer can use or monitor such high-risk 114
artificial intelligence system when such high-risk artificial intelligence 115
system is deployed. 116
(c) Except as provided in subsection (f) of this section, each developer 117
that offers, sells, leases, licenses, gives or otherwise makes available to a 118
deployer a high-risk artificial intelligence system on or after July 1, 2025, 119
shall provide to the deployer, as technically feasible, through artifacts 120
such as model cards, dataset cards or impact assessments, all material 121
information and documentation in the developer's possession, custody 122
or control that the deployer, or a third party contracted by the deployer, 123
may reasonably require to complete an impact assessment pursuant to 124
subsection (c) of section 3 of this act. 125
(d) (1) Beginning on July 1, 2025, each developer shall make available, 126
in a manner that is clear and readily available for public inspection on 127
such developer's Internet web site or in a public use case inventory, a 128
statement summarizing: 129
(A) The types of high-risk artificial intelligence systems that such 130
developer (i) has developed or intentionally and substantially modified, 131
and (ii) currently makes available to deployers; and 132
(B) How such developer manages known or reasonably foreseeable 133
risks of algorithmic discrimination arising from development or 134
intentional and substantial modification of the types of high-risk 135
artificial intelligence systems described in subparagraph (A) of this 136
Committee Bill No. 2
LCO No. 1489 6 of 37
subdivision. 137
(2) Each developer shall update the statement described in 138
subdivision (1) of this subsection (i) as necessary to ensure that such 139
statement remains accurate, and (ii) not later than ninety days after the 140
developer intentionally and substantially modifies any high-risk 141
artificial intelligence system described in subparagraph (A) of 142
subdivision (1) of this subsection. 143
(e) Beginning on July 1, 2025, if the developer of a high-risk artificial 144
intelligence system is informed by a deployer, or discovers through such 145
developer's ongoing testing and analysis, that such developer's high-146
risk artificial intelligence system has been deployed and caused, or is 147
reasonably likely to have caused, algorithmic discrimination, the 148
developer shall, not later than ninety days after the date of such 149
discovery, disclose to the Attorney General and all known deployers of 150
such high-risk artificial intelligence system any known or reasonably 151
foreseeable risk of algorithmic discrimination arising from the intended 152
uses of such high-risk artificial intelligence system. 153
(f) Nothing in subsections (b) to (e), inclusive, of this section shall be 154
construed to require a developer to disclose any trade secret or other 155
confidential or proprietary information. 156
(g) Beginning on July 1, 2025, the Attorney General may require that 157
a developer disclose to the Attorney General, in a form and manner 158
prescribed by the Attorney General, any statement or documentation 159
described in subsection (b) of this section if such statement or 160
documentation is relevant to an investigation conducted by the 161
Attorney General. The Attorney General may evaluate such statement 162
or documentation to ensure compliance with the provisions of this 163
section, and such statement or documentation shall be exempt from 164
disclosure under the Freedom of Information Act, as defined in section 165
1-200 of the general statutes. To the extent any information contained in 166
any such statement or documentation includes any information subject 167
to the attorney-client privilege or work product protection, such 168
Committee Bill No. 2
LCO No. 1489 7 of 37
disclosure shall not constitute a waiver of such privilege or protection. 169
Sec. 3. (NEW) (Effective October 1, 2024) (a) Beginning on July 1, 2025, 170
each deployer shall use reasonable care to protect consumers from any 171
known or reasonably foreseeable risks of algorithmic discrimination. In 172
any enforcement action brought on or after said date by the Attorney 173
General pursuant to section 7 of this act, there shall be a rebuttable 174
presumption that a deployer used reasonable care as required under 175
this subsection if the deployer complied with the provisions of 176
subsections (b) to (e), inclusive, of this section. 177
(b) (1) Beginning on July 1, 2025, no deployer shall deploy a high-risk 178
artificial intelligence system unless the deployer has implemented a risk 179
management policy and program. The risk management policy and 180
program shall specify and incorporate the principles, processes and 181
personnel that the deployer shall use to identify, document and 182
eliminate any known or reasonably foreseeable risks of algorithmic 183
discrimination. Each risk management policy and program 184
implemented and maintained pursuant to this subsection shall be 185
reasonable considering: 186
(A) (i) The guidance and standards set forth in the latest version of 187
the "Artificial Intelligence Risk Management Framework" published by 188
the National Institute of Standards and Technology or another 189
nationally or internationally recognized risk management framework 190
for artificial intelligence systems; or 191
(ii) Any risk management framework for artificial intelligence 192
systems that the Attorney General, in the Attorney General's discretion, 193
may designate; 194
(B) The size and complexity of the deployer; 195
(C) The nature and scope of the high-risk artificial intelligence 196
systems deployed by the deployer, including, but not limited to, the 197
intended uses of such high-risk artificial intelligence systems; and 198
Committee Bill No. 2
LCO No. 1489 8 of 37
(D) The sensitivity and volume of data processed in connection with 199
the high-risk artificial intelligence systems deployed by the deployer. 200
(2) A risk management policy and program implemented pursuant 201
to subdivision (1) of this subsection may cover multiple high-risk 202
artificial intelligence systems deployed by the deployer. 203
(c) (1) Except as provided in subdivisions (3) and (4) of this 204
subsection: 205
(A) A deployer that deploys a high-risk artificial intelligence system 206
on or after July 1, 2025, or a third party contracted by the deployer, shall 207
complete an impact assessment for the high-risk artificial intelligence 208
system; and 209
(B) Beginning on July 1, 2025, a deployer, or a third party contracted 210
by the deployer, shall complete an impact assessment for a deployed 211
high-risk artificial intelligence system not later than ninety days after 212
any intentional and substantial modification to such high-risk artificial 213
intelligence system is made available. 214
(2) (A) Each impact assessment completed pursuant to this subsection 215
shall include, at a minimum: 216
(i) A statement by the deployer disclosing the purpose, intended use 217
cases and deployment context of, and benefits afforded by, the high-risk 218
artificial intelligence system; 219
(ii) An analysis of whether the deployment of the high-risk artificial 220
intelligence system poses any known or reasonably foreseeable risks of 221
algorithmic discrimination and, if so, the nature of such algorithmic 222
discrimination and the steps that have been taken to eliminate such 223
risks; 224
(iii) A description of (I) the categories of data the high-risk artificial 225
intelligence system processes as inputs, and (II) the outputs such high-226
risk artificial intelligence system produces; 227
Committee Bill No. 2
LCO No. 1489 9 of 37
(iv) If the deployer used data to customize the high-risk artificial 228
intelligence system, an overview of the categories of data the deployer 229
used to retrain such high-risk artificial intelligence system; 230
(v) Any metrics used to evaluate the performance and known 231
limitations of the high-risk artificial intelligence system; 232
(vi) A description of any transparency measures taken concerning the 233
high-risk artificial intelligence system, including, but not limited to, any 234
measures taken to disclose to a consumer that such high-risk artificial 235
intelligence system is in use when such high-risk artificial intelligence 236
system is in use; and 237
(vii) A description of the post-deployment monitoring and user 238
safeguards provided concerning such high-risk artificial intelligence 239
system, including, but not limited to, the oversight process established 240
by the deployer to address issues arising from deployment of such high-241
risk artificial intelligence system as such issues arise. 242
(B) In addition to the statement, analysis, descriptions, overview and 243
metrics required under subparagraph (A) of this subdivision, each 244
impact assessment completed pursuant to this subsection following an 245
intentional and substantial modification made to a high-risk artificial 246
intelligence system on or after July 1, 2025, shall include a statement 247
disclosing the extent to which the high-risk artificial intelligence system 248
was used in a manner that was consistent with, or varied from, the 249
developer's intended uses of such high-risk artificial intelligence 250
system. 251
(3) A single impact assessment may address a comparable set of high-252
risk artificial intelligence systems deployed by a deployer. 253
(4) If a deployer, or a third party contracted by the deployer, 254
completes an impact assessment for the purpose of complying with 255
another applicable law or regulation, such impact assessment shall be 256
deemed to satisfy the requirements established in this subsection if such 257
Committee Bill No. 2
LCO No. 1489 10 of 37
impact assessment is reasonably similar in scope and effect to the impact 258
assessment that would otherwise be completed pursuant to this 259
subsection. 260
(5) A deployer shall maintain the most recently completed impact 261
assessment for a high-risk artificial intelligence system as required 262
under this subsection, all records concerning each such impact 263
assessment and all prior impact assessments, if any, for a period of at 264
least three years following the final deployment of the high-risk artificial 265
intelligence system. 266
(d) Beginning on July 1, 2025, a deployer, or a third party contracted 267
by the deployer, shall review, at least annually, the deployment of each 268
high-risk artificial intelligence system deployed by the deployer to 269
ensure that such high-risk artificial intelligence system is not causing 270
algorithmic discrimination. 271
(e) (1) Beginning on July 1, 2025, and not later than the time that a 272
deployer deploys a high-risk artificial intelligence system to make, or be 273
a controlling factor in making, a consequential decision concerning a 274
consumer, the deployer shall: 275
(A) Notify the consumer that the deployer has deployed a high-risk 276
artificial intelligence system to make, or be a controlling factor in 277
making, such consequential decision; and 278
(B) Provide to the consumer (i) a statement disclosing (I) the purpose 279
of such high-risk artificial intelligence system, and (II) the nature of such 280
consequential decision, (ii) contact information for such deployer, and 281
(iii) a description, in plain language, of such high-risk artificial 282
intelligence system, which description shall, at a minimum, include a 283
description of (I) any human components of such high-risk artificial 284
intelligence system, and (II) how any automated components of such 285
high-risk artificial intelligence system are used to inform such 286
consequential decision. 287
Committee Bill No. 2
LCO No. 1489 11 of 37
(2) A deployer may provide to a consumer the notice, statement, 288
contact information and description required under subdivision (1) of 289
this subsection in any manner that is clear and readily available. 290
(f) (1) Beginning on July 1, 2025, each deployer shall make available, 291
in a manner that is clear and readily available for public inspection, a 292
statement summarizing: 293
(A) The types of high-risk artificial intelligence systems that are 294
currently deployed by such deployer; and 295
(B) How such deployer manages any known or reasonably 296
foreseeable risks of algorithmic discrimination that may arise from 297
deployment of each high-risk artificial intelligence system described in 298
subparagraph (A) of this subdivision. 299
(2) Each deployer shall periodically update the statement described 300
in subdivision (1) of this subsection. 301
(g) If a deployer deploys a high-risk artificial intelligence system on 302
or after July 1, 2025, and subsequently discovers that the high-risk 303
artificial intelligence system has caused, or is reasonably likely to have 304
caused, algorithmic discrimination against consumers, the deployer 305
shall, not later than ninety days after the date of such discovery, send to 306
the Attorney General, in a form and manner prescribed by the Attorney 307
General, a notice disclosing such discovery. 308
(h) Nothing in subsections (b) to (g), inclusive, of this section shall be 309
construed to require a deployer to disclose any trade secret. 310
(i) Beginning on July 1, 2025, the Attorney General may require that 311
a deployer, or the third party contracted by the deployer as set forth in 312
subsection (c) of this section, as applicable, disclose to the Attorney 313
General, in a form and manner prescribed by the Attorney General, any 314
risk management policy implemented pursuant to subsection (b) of this 315
section, impact assessment completed pursuant to subsection (c) of this 316
section or record maintained pursuant to subdivision (5) of subsection 317
Committee Bill No. 2
LCO No. 1489 12 of 37
(c) of this section if such risk management policy, impact assessment or 318
record is relevant to an investigation conducted by the Attorney 319
General. The Attorney General may evaluate such risk management 320
policy, impact assessment or record to ensure compliance with the 321
provisions of this section, and such risk management policy, impact 322
assessment or record shall be exempt from disclosure under the 323
Freedom of Information Act, as defined in section 1-200 of the general 324
statutes. To the extent any information contained in any such risk 325
management policy, impact assessment or record includes any 326
information subject to the attorney-client privilege or work product 327
protection, such disclosure shall not constitute a waiver of such 328
privilege or protection. 329
Sec. 4. (NEW) (Effective October 1, 2024) (a) (1) Beginning on January 330
1, 2026, each developer shall use reasonable care to protect consumers 331
from any risk arising from any development or intentional and 332
substantial modification of a generative artificial intelligence system, to 333
the extent such risk is known or reasonably foreseeable: 334
(A) Of any unfair or deceptive trade practice under subsection (a) of 335
section 42-110b of the general statutes; 336
(B) Of any unlawful disparate impact on consumers; 337
(C) Of any emotional, financial, mental, physical or reputational 338
injury to consumers that may be redressed under the laws of this state; 339
(D) Of any physical or other intrusion upon the solitude or seclusion, 340
or the private affairs or concerns, of consumers if such intrusion (i) 341
would be offensive to a reasonable person, and (ii) may be redressed 342
under the laws of this state; or 343
(E) To the intellectual property rights of persons under applicable 344
state and federal intellectual property laws. 345
(2) In any enforcement action brought by the Attorney General 346
pursuant to section 7 of this act on or after January 1, 2026, there shall 347
Committee Bill No. 2
LCO No. 1489 13 of 37
be a rebuttable presumption that a developer used reasonable care as 348
required under subdivision (1) of this subsection if the developer 349
complied with the provisions of this section. 350
(b) (1) Except as provided in subdivision (2) of this subsection, a 351
developer that develops, or intentionally and substantially modifies, a 352
general purpose artificial intelligence model on or after January 1, 2026, 353
shall: 354
(A) Reduce and mitigate the known or reasonably foreseeable risks 355
described in subdivision (1) of subsection (a) of this section through, for 356
example, the involvement of qualified experts and documentation of 357
any known or reasonably foreseeable, but nonmitigable, risks; 358
(B) Incorporate and process datasets that are subject to data 359
governance measures, including, but not limited to, measures to (i) 360
examine the suitability of data sources for possible biases and 361
appropriate mitigation, and (ii) prevent such general purpose artificial 362
intelligence model from recklessly training on child pornography, as 363
defined in section 53a-193 of the general statutes; 364
(C) Achieve, throughout the lifecycle of such general purpose 365
artificial intelligence model, appropriate levels of performance, 366
predictability, interpretability, corrigibility, safety and cybersecurity, as 367
assessed through appropriate methods, including, but not limited to, 368
model evaluation involving independent experts, documented analysis 369
and extensive testing, during conceptualization, design and 370
development of such general purpose artificial intelligence model; and 371
(D) Incorporate science-backed standards and techniques that (i) 372
authenticate, detect, label and track the provenance of audio files, 373
images or videos that are synthetic digital content, where appropriate, 374
and in a manner that is (I) technically feasible, and (II) informed by the 375
specificities and limitations of different content types, and (ii) ensure 376
that such general purpose artificial intelligence model includes 377
safeguards that are (I) adequate to prevent generation of content in 378
Committee Bill No. 2
LCO No. 1489 14 of 37
violation of applicable law, including, but not limited to, child 379
pornography, as defined in section 53a-193 of the general statutes, and 380
(II) in line with the generally acknowledged state of the art; and 381
(2) (A) The provisions of subdivision (1) of this subsection shall not 382
apply to a developer that develops, or intentionally and substantially 383
modifies, a general purpose artificial intelligence model on or after 384
January 1, 2026, if: 385
(i) The developer releases such general purpose artificial intelligence 386
model under a free and open-source license; and 387
(ii) Unless such general purpose artificial intelligence model is 388
deployed as a high-risk artificial intelligence system, the parameters of 389
such general purpose artificial intelligence model, including, but not 390
limited to, the weights and information concerning the model 391
architecture and model usage for such general purpose artificial 392
intelligence model, are made publicly available. 393
(B) A developer that takes any action under the exemption 394
established in subparagraph (A) of this subdivision shall bear the 395
burden of demonstrating that such action qualifies for such exemption. 396
(3) A developer that develops, or intentionally and substantially 397
modifies, a general purpose artificial intelligence model described in 398
subdivision (1) of this subsection shall maintain all records maintained 399
for the purposes set forth in this subsection for a period of at least three 400
years following the final deployment of such general purpose artificial 401
intelligence model. 402
(c) (1) Except as provided in subdivisions (3) and (4) of this 403
subsection, a developer that develops, or intentionally and substantially 404
modifies, a generative artificial intelligence system on or after January 405
1, 2026, shall complete an impact assessment for such generative 406
artificial intelligence system pursuant to this subsection. 407
(2) Each impact assessment completed pursuant to this subsection 408
Committee Bill No. 2
LCO No. 1489 15 of 37
shall include, at a minimum, an evaluation of: 409
(A) The intended purpose and potential benefits of such generative 410
artificial intelligence system; 411
(B) Any reasonably foreseeable risk that such generative artificial 412
intelligence system could (i) harm the health or safety of individuals, or 413
(ii) result in unlawful discrimination against individuals; 414
(C) Whether use of such generative artificial intelligence system 415
could harm the health and safety of individuals or adversely impact the 416
fundamental rights of individuals; and 417
(D) The extent to which individuals who may be harmed or adversely 418
impacted are dependent on the outcomes produced by such generative 419
artificial intelligence system. 420
(3) A single impact assessment may address a comparable set of 421
generative artificial intelligence systems developed, or intentionally and 422
substantially modified, by a developer. 423
(4) If a developer completes an impact assessment for the purpose of 424
complying with another applicable law or regulation, such impact 425
assessment shall be deemed to satisfy the requirements established in 426
this subsection if such impact assessment is reasonably similar in scope 427
and effect to the impact assessment that would otherwise be completed 428
pursuant to this subsection. 429
(5) A developer that completes an impact assessment pursuant to this 430
subsection shall maintain such impact assessment, and all records 431
concerning such impact assessment, for a period of at least three years 432
following the final deployment of such generative artificial intelligence 433
system. 434
(d) Beginning on January 1, 2026, the Attorney General may require 435
that a developer disclose to the Attorney General, in a form and manner 436
prescribed by the Attorney General, any record maintained pursuant to 437
Committee Bill No. 2
LCO No. 1489 16 of 37
subdivision (3) of subsection (b) of this section, impact assessment 438
completed pursuant to subsection (c) of this section or record 439
maintained pursuant to subdivision (5) of subsection (c) of this section 440
if such impact assessment or record is relevant to an investigation 441
conducted by the Attorney General. The Attorney General may evaluate 442
such impact assessment or record to ensure compliance with the 443
provisions of this section, and such impact assessment or record shall be 444
exempt from disclosure under the Freedom of Information Act, as 445
defined in section 1-200 of the general statutes. To the extent any 446
information contained in any such impact assessment or record includes 447
any information subject to the attorney-client privilege or work product 448
protection, such disclosure shall not constitute a waiver of such 449
privilege or protection. 450
Sec. 5. (NEW) (Effective October 1, 2024) (a) Except as provided in 451
subsection (b) of this section, each person doing business in this state, 452
including, but not limited to, each developer or deployer that develops, 453
intentionally and substantially modifies, deploys, offers, sells, leases, 454
licenses, gives or otherwise provides, as applicable, an artificial 455
intelligence system that is intended to interact with consumers shall 456
ensure that such artificial intelligence system discloses to each consumer 457
who interacts with such artificial intelligence system that such consumer 458
is interacting with an artificial intelligence system. 459
(b) No disclosure shall be required under subsection (a) of this section 460
under circumstances in which: 461
(1) A reasonable person would deem it obvious that such person is 462
interacting with an artificial intelligence system; or 463
(2) The developer or deployer did not directly make the artificial 464
intelligence system available to consumers. 465
Sec. 6. (NEW) (Effective October 1, 2024) (a) Nothing in sections 1 to 7, 466
inclusive, of this act shall be construed to restrict a developer's or 467
deployer's ability to: (1) Comply with federal, state or municipal 468
Committee Bill No. 2
LCO No. 1489 17 of 37
ordinances or regulations; (2) comply with a civil, criminal or regulatory 469
inquiry, investigation, subpoena or summons by federal, state, 470
municipal or other governmental authorities; (3) cooperate with law 471
enforcement agencies concerning conduct or activity that the developer 472
or deployer reasonably and in good faith believes may violate federal, 473
state or municipal ordinances or regulations; (4) investigate, establish, 474
exercise, prepare for or defend legal claims; (5) take immediate steps to 475
protect an interest that is essential for the life or physical safety of the 476
consumer or another individual; (6) prevent, detect, protect against or 477
respond to security incidents, identity theft, fraud, harassment, 478
malicious or deceptive activities or any illegal activity, preserve the 479
integrity or security of systems or investigate, report or prosecute those 480
responsible for any such action; (7) engage in public or peer-reviewed 481
scientific or statistical research in the public interest that adheres to all 482
other applicable ethics and privacy laws and is approved, monitored 483
and governed by an institutional review board that determines, or by 484
similar independent oversight entities that determine, (A) that the 485
expected benefits of the research outweigh the risks associated with 486
such research, and (B) whether the developer or deployer has 487
implemented reasonable safeguards to mitigate the risks associated 488
with such research; or (8) assist another developer or deployer with any 489
of the obligations imposed under sections 1 to 7, inclusive, of this act. 490
(b) The obligations imposed on developers or deployers under 491
sections 1 to 7, inclusive, of this act shall not restrict a developer's or 492
deployer's ability to: (1) Effectuate a product recall; or (2) identify and 493
repair technical errors that impair existing or intended functionality. 494
(c) The obligations imposed on developers or deployers under 495
sections 1 to 7, inclusive, of this act shall not apply where compliance by 496
the developer or deployer with said sections would violate an 497
evidentiary privilege under the laws of this state. 498
(d) Nothing in sections 1 to 7, inclusive, of this act shall be construed 499
to impose any obligation on a developer or deployer that adversely 500
Committee Bill No. 2
LCO No. 1489 18 of 37
affects the rights or freedoms of any person, including, but not limited 501
to, the rights of any person: (1) To freedom of speech or freedom of the 502
press guaranteed in the First Amendment to the United States 503
Constitution; or (2) under section 52-146t of the general statutes. 504
(e) If a developer or deployer engages in any action pursuant to an 505
exemption set forth in subsections (a) to (d), inclusive, of this section, 506
the developer or deployer bears the burden of demonstrating that such 507
action qualifies for such exemption. 508
Sec. 7. (NEW) (Effective October 1, 2024) (a) The Attorney General shall 509
have exclusive authority to enforce the provisions of sections 1 to 6, 510
inclusive, of this act. 511
(b) Except as provided in subsection (f) of this section, during the 512
period beginning on July 1, 2025, and ending on June 30, 2026, the 513
Attorney General shall, prior to initiating any action for a violation of 514
any provision of sections 1 to 6, inclusive, of this act, issue a notice of 515
violation to the developer or deployer if the Attorney General 516
determines that it is possible to cure such violation. If the developer or 517
deployer fails to cure such violation not later than sixty days after 518
receipt of the notice of violation, the Attorney General may bring an 519
action pursuant to this section. Not later than January 1, 2027, the 520
Attorney General shall submit a report, in accordance with section 11-521
4a of the general statutes, to the joint standing committee of the General 522
Assembly having cognizance of matters relating to consumer protection 523
disclosing: (1) The number of notices of violation the Attorney General 524
has issued; (2) the nature of each violation; (3) the number of violations 525
that were cured during the sixty-day cure period; and (4) any other 526
matter the Attorney General deems relevant for the purposes of such 527
report. 528
(c) Except as provided in subsection (f) of this section, beginning on 529
July 1, 2026, the Attorney General may, in determining whether to grant 530
a developer or deployer the opportunity to cure an alleged violation 531
described in subsection (b) of this section, consider: (1) The number of 532
Committee Bill No. 2
LCO No. 1489 19 of 37
violations; (2) the size and complexity of the developer or deployer; (3) 533
the nature and extent of the developer's or deployer's business; (4) the 534
substantial likelihood of injury to the public; (5) the safety of persons or 535
property; and (6) whether such alleged violation was likely caused by 536
human or technical error. 537
(d) Nothing in sections 1 to 6, inclusive, of this act shall be construed 538
as providing the basis for a private right of action for violations of said 539
sections or any other law. 540
(e) Except as provided in subsection (f) of this section, a violation of 541
the requirements established in sections 1 to 6, inclusive, of this act shall 542
constitute an unfair trade practice for purposes of section 42-110b of the 543
general statutes and shall be enforced solely by the Attorney General, 544
provided the provisions of section 42-110g of the general statutes shall 545
not apply to such violation. 546
(f) (1) In any action commenced by the Attorney General for any 547
violation of sections 1 to 6, inclusive, of this act, it shall be an affirmative 548
defense that: 549
(A) The developer or deployer of the generative artificial intelligence 550
system or high-risk artificial intelligence system, as applicable, 551
implemented and maintains a program that is in compliance with: 552
(i) The latest version of the "Artificial Intelligence Risk Management 553
Framework" published by the National Institute of Standards and 554
Technology or another nationally or internationally recognized risk 555
management framework for artificial intelligence systems; 556
(ii) Any risk management framework for artificial intelligence 557
systems that the Attorney General, in the Attorney General's discretion, 558
may designate; or 559
(iii) Any risk management framework for artificial intelligence 560
systems designated by the Banking Commissioner or Insurance 561
Commissioner if the developer or deployer is regulated by the 562
Committee Bill No. 2
LCO No. 1489 20 of 37
Department of Banking or Insurance Department; and 563
(B) The developer or deployer: 564
(i) Encourages the deployers or users of the generative artificial 565
intelligence system or high-risk artificial intelligence system, as 566
applicable, to provide feedback to such developer or deployer; 567
(ii) Discovers a violation of any provision of sections 1 to 6, inclusive, 568
of this act (I) as a result of the feedback described in subparagraph (B)(i) 569
of this subdivision, (II) through adversarial testing or red-teaming, as 570
such terms are defined or used by the National Institutes of Standards 571
and Technology, or (III) through an internal review process; and 572
(iii) Not later than sixty days after discovering the violation as set 573
forth in subparagraph (B)(ii) of this subdivision, (I) cures such violation, 574
and (II) provides to the Attorney General, in a form and manner 575
prescribed by the Attorney General, notice that such violation has been 576
cured and evidence that any harm caused by such violation has been 577
mitigated. 578
(2) The developer or deployer bears the burden of demonstrating to 579
the Attorney General that the requirements established in subdivision 580
(1) of this subsection have been satisfied. 581
Sec. 8. (NEW) (Effective from passage) (a) For the purposes of this 582
section, "artificial intelligence" means: (1) An artificial system that (A) 583
performs tasks under varying and unpredictable circumstances without 584
significant human oversight or can learn from experience and improve 585
such performance when exposed to datasets, (B) is developed in any 586
context, including, but not limited to, software or physical hardware, 587
and solves tasks requiring human-like perception, cognition, planning, 588
learning, communication or physical action, or (C) is designed to (i) 589
think or act like a human by using, for example, a cognitive architecture 590
or neural network, or (ii) act rationally by using, for example, an 591
intelligent software agent or embodied robot that achieves goals 592
Committee Bill No. 2
LCO No. 1489 21 of 37
through perception, planning, reasoning, learning, communication, 593
decision-making or action; and (2) a set of techniques, including, but not 594
limited to, machine learning, that is designed to approximate a cognitive 595
task. 596
(b) There is established an Artificial Intelligence Advisory Council to 597
engage stakeholders and experts to: (1) Make recommendations 598
concerning, and develop best practices for, the ethical and equitable use 599
of artificial intelligence in state government; (2) assess the White House 600
Office of Science and Technology Policy's "Blueprint for an AI Bill of 601
Rights" and similar materials and make recommendations concerning 602
the (A) regulation of the use of artificial intelligence in the private sector 603
based, among other things, on said blueprint, and (B) adoption of a 604
Connecticut artificial intelligence bill of rights based on said blueprint; 605
and (3) make recommendations concerning the adoption of other 606
legislation concerning artificial intelligence. 607
(c) (1) (A) The advisory council shall be part of the Legislative 608
Department and consist of the following voting members: (i) One 609
appointed by the speaker of the House of Representatives, who shall be 610
a representative of the industries that are developing artificial 611
intelligence; (ii) one appointed by the president pro tempore of the 612
Senate, who shall be a representative of the industries that are using 613
artificial intelligence; (iii) one appointed by the majority leader of the 614
House of Representatives, who shall be an academic with a 615
concentration in the study of technology and technology policy; (iv) one 616
appointed by the majority leader of the Senate, who shall be an academic 617
with a concentration in the study of government and public policy; (v) 618
one appointed by the minority leader of the House of Representatives, 619
who shall be a representative of an industry association representing the 620
industries that are developing artificial intelligence; (vi) one appointed 621
by the minority leader of the Senate, who shall be a representative of an 622
industry association representing the industries that are using artificial 623
intelligence; (vii) one appointed by the House chairperson of the joint 624
standing committee of the General Assembly having cognizance of 625
Committee Bill No. 2
LCO No. 1489 22 of 37
matters relating to consumer protection; (viii) one appointed by the 626
Senate chairperson of the joint standing committee of the General 627
Assembly having cognizance of matters relating to consumer 628
protection; (ix) two appointed by the Governor, who shall be members 629
of the Connecticut Academy of Science and Engineering; and (x) the 630
House and Senate chairpersons of the joint standing committee of the 631
General Assembly having cognizance of matters relating to consumer 632
protection. 633
(B) All voting members appointed pursuant to subparagraphs (A)(i) 634
to (A)(ix), inclusive, of this subdivision shall have professional 635
experience or academic qualifications in matters pertaining to artificial 636
intelligence, automated systems, government policy or another related 637
field. 638
(C) All initial appointments to the advisory council under 639
subparagraphs (A)(i) to (A)(ix), inclusive, of this subdivision shall be 640
made not later than thirty days after the effective date of this section. 641
Any vacancy shall be filled by the appointing authority. 642
(D) Any action taken by the advisory council shall be taken by a 643
majority vote of all members present who are entitled to vote, provided 644
no such action may be taken unless at least fifty per cent of such 645
members are present. 646
(2) The advisory council shall include the following nonvoting, ex-647
officio members: (A) The Attorney General, or the Attorney General's 648
designee; (B) the Comptroller, or the Comptroller's designee; (C) the 649
Treasurer, or the Treasurer's designee; (D) the Commissioner of 650
Administrative Services, or said commissioner's designee; (E) the Chief 651
Data Officer, or said officer's designee; (F) the executive director of the 652
Freedom of Information Commission, or said executive director's 653
designee; (G) the executive director of the Commission on Women, 654
Children, Seniors, Equity and Opportunity, or said executive director's 655
designee; (H) the Chief Court Administrator, or said administrator's 656
designee; and (I) the executive director of the Connecticut Academy of 657
Committee Bill No. 2
LCO No. 1489 23 of 37
Science and Engineering, or said executive director's designee. 658
(d) The chairpersons of the joint standing committee of the General 659
Assembly having cognizance of matters relating to consumer protection 660
shall serve as chairpersons of the advisory council. Such chairpersons 661
shall schedule the first meeting of the advisory council, which shall be 662
held not later than sixty days after the effective date of this section. 663
(e) The administrative staff of the joint standing committee of the 664
General Assembly having cognizance of matters relating to consumer 665
protection shall serve as administrative staff of the advisory council. 666
Sec. 9. Subsection (a) of section 53a-189c of the general statutes is 667
repealed and the following is substituted in lieu thereof (Effective October 668
1, 2024): 669
(a) A person is guilty of unlawful dissemination of an intimate image 670
when (1) such person intentionally disseminates by electronic or other 671
means a photograph, film, videotape or other recorded [image] or 672
synthetic image of (A) the genitals, pubic area or buttocks of another 673
person with less than a fully opaque covering of such body part, or the 674
breast of such other person who is female with less than a fully opaque 675
covering of any portion of such breast below the top of the nipple, or (B) 676
another person engaged in sexual intercourse, as defined in section 53a-677
193, (2) such person disseminates such image without the consent of 678
such other person, knowing that such other person understood that the 679
image would not be so disseminated, and (3) such other person suffers 680
harm as a result of such dissemination. For purposes of this subsection, 681
"disseminate" means to sell, give, provide, lend, trade, mail, deliver, 682
transfer, publish, distribute, circulate, present, exhibit, advertise or 683
otherwise offer; [, and] "harm" includes, but is not limited to, subjecting 684
such other person to hatred, contempt, ridicule, physical injury, 685
financial injury, psychological harm or serious emotional distress; and 686
"synthetic image" means an image that is partially or fully generated by 687
a computer system, and not wholly recorded by a camera. 688
Committee Bill No. 2
LCO No. 1489 24 of 37
Sec. 10. Section 9-600 of the general statutes is repealed and the 689
following is substituted in lieu thereof (Effective July 1, 2024): 690
[This] Except as otherwise provided in section 11 of this act, this 691
chapter applies to: (1) The election, and all primaries preliminary 692
thereto, of all public officials, except presidential electors, United States 693
senators and members in Congress, and (2) any referendum question. 694
This chapter also applies, except for the provisions of sections 9-611 to 695
9-620, inclusive, to persons who are candidates in a primary for town 696
committee members. 697
Sec. 11. (NEW) (Effective July 1, 2024) (a) As used in this section: 698
(1) "Artificial intelligence" means a machine-based system that (A) 699
can, for a given set of human-defined objectives, make predictions, 700
recommendations or decisions influencing real or virtual environments, 701
and (B) uses machine and human-based inputs to (i) perceive real and 702
virtual environments, (ii) abstract such perceptions into models through 703
analysis in an automated manner, and (iii) formulate options for 704
information or action through model inference; 705
(2) "Candidate" means a human being who seeks election, or 706
nomination for election, to any municipal, federal or state office; 707
(3) "Deceptive media" means an image, audio or video that (A) 708
depicts a human being engaging in speech or conduct in which the 709
human being did not engage, (B) a reasonable viewer or listener would 710
incorrectly believe depicts such human being engaging in such speech 711
or conduct, and (C) was produced, in whole or in part, by artificial 712
intelligence; 713
(4) "Election" has the same meaning as provided in section 9-1 of the 714
general statutes; and 715
(5) "Elector" has the same meaning as provided in section 9-1 of the 716
general statutes. 717
Committee Bill No. 2
LCO No. 1489 25 of 37
(b) Except as provided in subsection (c) of this section, no person shall 718
distribute, or enter into an agreement with another person to distribute, 719
any deceptive media during the ninety-day period preceding an 720
election, or any primary precedent thereto, if: 721
(1) The person (A) knows such deceptive media depicts any human 722
being engaging in speech or conduct in which such person did not 723
engage, and (B) in distributing such deceptive media or entering into 724
such agreement, intends to (i) harm the reputation or electoral prospects 725
of a candidate in the primary or election, and (ii) change the voting 726
behavior of electors in the primary or election by deceiving such electors 727
into incorrectly believing that the human being described in 728
subparagraph (A) of this subdivision engaged in the speech or conduct 729
described in said subparagraph; and 730
(2) It is reasonably foreseeable that the distribution will (A) harm the 731
reputation or electoral prospects of a candidate in the primary or 732
election, and (B) change the voting behavior of electors in the primary 733
or election in the manner set forth in subparagraph (B)(ii) of subdivision 734
(1) of this subsection. 735
(c) A person may distribute, or enter into an agreement with another 736
person to distribute, deceptive media during the ninety-day period 737
preceding a primary or election if the deceptive media includes a 738
disclaimer: 739
(1) Informing viewers or listeners, as applicable, that the media has 740
been manipulated by technical means and depicts speech or conduct 741
that did not occur; 742
(2) If the deceptive media is a video, that (A) appears throughout the 743
entirety of the video, (B) is clearly visible to, and readable by, the 744
average viewer, (C) is in letters (i) at least as large as the majority of the 745
other text included in the video, or (ii) if there is no other text included 746
in the video, in a size that is easily readable by the average viewer, and 747
(D) is in the same language otherwise used in such deceptive media; 748
Committee Bill No. 2
LCO No. 1489 26 of 37
(3) If the deceptive media exclusively consists of audio, that is read 749
(A) at the beginning and end of the media, (B) in a clearly spoken 750
manner, (C) in a pitch that can be easily heard by the average listener, 751
and (D) in the same language otherwise used in such deceptive media; 752
(4) If the deceptive media is an image, that (A) is clearly visible to, 753
and readable by, the average viewer, (B) if the media contains other text, 754
is in letters (i) at least as large as the majority of the other text included 755
in the image, or (ii) if there is no other text included in the image, in a 756
size that is easily readable by the average viewer, and (C) is in the same 757
language otherwise used in such deceptive media; and 758
(5) If the deceptive media was generated by editing an existing image, 759
audio or video, that includes a citation directing the viewer or listener 760
to the original source from which the unedited version of such existing 761
image, audio or video was obtained. 762
(d) (1) Any person who violates any provision of this section shall be 763
guilty of a class C misdemeanor, except that any violation committed 764
not later than five years after conviction for a prior violation shall be a 765
class D felony. 766
(2) Any penalty imposed under subdivision (1) of this subsection 767
shall be in addition to any injunctive or other equitable relief ordered 768
under subsection (e) of this section. 769
(e) (1) The Attorney General, a human being described in 770
subparagraph (A) of subdivision (1) of subsection (b) of this section or 771
candidate for office who has been, or is likely to be, injured by the 772
distribution of deceptive media in violation of the provisions of this 773
section, or an organization that represents the interests of electors who 774
have been, or are likely to be, deceived by any such distribution, may 775
commence a civil action, in a court of competent jurisdiction, seeking to 776
permanently enjoin any person who is alleged to have committed such 777
violation from continuing such violation. No court shall have 778
jurisdiction to grant extraordinary relief in the form of a temporary 779
Committee Bill No. 2
LCO No. 1489 27 of 37
restraining order or preliminary injunction for any violation of this 780
section. 781
(2) In any civil action commenced under subdivision (1) of this 782
subsection, the plaintiff shall bear the burden of proving, by clear and 783
convincing evidence, that the defendant distributed deceptive media in 784
violation of the provisions of this section. 785
(3) Any party, other than the Attorney General, who prevails in a civil 786
action commenced under subdivision (1) of this subsection shall be 787
awarded reasonable attorney's fees and costs to be taxed by the court. 788
Sec. 12. (Effective from passage) (a) As used in this section: 789
(1) "Artificial intelligence" means any technology, including, but not 790
limited to, machine learning, that uses data to train an algorithm or 791
predictive model for the purpose of enabling a computer system or 792
service to autonomously perform any task, including, but not limited to, 793
visual perception, language processing or speech recognition, that is 794
normally associated with human intelligence or perception; 795
(2) "Generative artificial intelligence" means any form of artificial 796
intelligence, including, but not limited to, a foundation model, that is 797
able to produce synthetic digital content; 798
(3) "Machine learning" means any technique that enables a computer 799
system or service to autonomously learn and adapt by using algorithms 800
and statistical models to autonomously analyze and draw inferences 801
from patterns in data; and 802
(4) "State agency" means any department, board, council, 803
commission, institution or other executive branch agency of state 804
government, including, but not limited to, each constituent unit and 805
each public institution of higher education. 806
(b) Each state agency shall study how generative artificial intelligence 807
may be incorporated in its processes to improve efficiencies. Each state 808
Committee Bill No. 2
LCO No. 1489 28 of 37
agency shall solicit input from its employees concerning such 809
incorporation, including, but not limited to, any applicable collective 810
bargaining unit that represents its employees and appropriate experts 811
from civil society organizations, academia and industry. 812
(c) Not later than January 1, 2025, each state agency shall submit the 813
results of such study to the Department of Administrative Services, 814
including a request for approval of any potential pilot project utilizing 815
generative artificial intelligence that the state agency intends to 816
establish, provided such use is in accordance with the policies and 817
procedures established by the Office of Policy and Management 818
pursuant to subsection (b) of section 4-68jj of the general statutes. Any 819
such pilot project shall measure how generative artificial intelligence (1) 820
improves Connecticut residents' experience with and access to 821
government services, and (2) supports state agency employees in the 822
performance of their duties in addition to any domain-specific impacts 823
to be measured by the state agency. The Commissioner of 824
Administrative Services shall assess any such proposed pilot project in 825
accordance with the provisions of section 4a-2e of the general statutes, 826
as amended by this act, and may disapprove any pilot project that fails 827
such assessment or requires additional legislative authorization. 828
(d) Not later than February 1, 2025, the Commissioner of 829
Administrative Services shall submit a report, in accordance with the 830
provisions of section 11-4a of the general statutes, to the joint standing 831
committees of the General Assembly having cognizance of matters 832
relating to consumer protection and government administration. Such 833
report shall include a summary of all pilot projects approved by the 834
commissioner under this section and any recommendations for 835
legislation necessary to implement additional pilot projects. 836
Sec. 13. Section 4a-2e of the 2024 supplement to the general statutes 837
is repealed and the following is substituted in lieu thereof (Effective July 838
1, 2024): 839
(a) For the purposes of this section: 840
Committee Bill No. 2
LCO No. 1489 29 of 37
(1) "Artificial intelligence" means (A) an artificial system that (i) 841
performs tasks under varying and unpredictable circumstances without 842
significant human oversight or can learn from experience and improve 843
such performance when exposed to data sets, (ii) is developed in any 844
context, including, but not limited to, software or physical hardware, 845
and solves tasks requiring human-like perception, cognition, planning, 846
learning, communication or physical action, or (iii) is designed to (I) 847
think or act like a human, including, but not limited to, a cognitive 848
architecture or neural network, or (II) act rationally, including, but not 849
limited to, an intelligent software agent or embodied robot that achieves 850
goals using perception, planning, reasoning, learning, communication, 851
decision-making or action, or (B) a set of techniques, including, but not 852
limited to, machine learning, that is designed to approximate a cognitive 853
task; [and] 854
(2) "Generative artificial intelligence" means any form of artificial 855
intelligence, including, but not limited to, a foundation model, that is 856
able to produce synthetic digital content; and 857
[(2)] (3) "State agency" has the same meaning as provided in section 858
4d-1. 859
(b) (1) Not later than December 31, 2023, and annually thereafter, the 860
[Department] Commissioner of Administrative Services shall conduct 861
an inventory of all systems that employ artificial intelligence and are in 862
use by any state agency. Each such inventory shall include at least the 863
following information for each such system: 864
(A) The name of such system and the vendor, if any, that provided 865
such system; 866
(B) A description of the general capabilities and uses of such system; 867
(C) Whether such system was used to independently make, inform or 868
materially support a conclusion, decision or judgment; and 869
(D) Whether such system underwent an impact assessment prior to 870
Committee Bill No. 2
LCO No. 1489 30 of 37
implementation. 871
(2) The [Department] Commissioner of Administrative Services shall 872
make each inventory conducted pursuant to subdivision (1) of this 873
subsection publicly available on the state's open data portal. 874
(c) Beginning on February 1, 2024, the [Department] Commissioner 875
of Administrative Services shall perform ongoing assessments of 876
systems that employ artificial intelligence and are in use by state 877
agencies to ensure that no such system shall result in any unlawful 878
discrimination or disparate impact described in subparagraph (B) of 879
subdivision (1) of subsection (b) of section 4-68jj. The [department] 880
commissioner shall perform such assessment in accordance with the 881
policies and procedures established by the Office of Policy and 882
Management pursuant to subsection (b) of section 4-68jj. 883
(d) The Commissioner of Administrative Services shall, in 884
consultation with other state agencies, collective bargaining units that 885
represent state agency employees and industry experts, develop 886
trainings for state agency employees on (1) the use of generative 887
artificial intelligence tools that are determined by the commissioner, 888
pursuant to the assessment performed under subsection (c) of this 889
section, to achieve equitable outcomes, and (2) methods for identifying 890
and mitigating potential output inaccuracies, fabricated text, 891
hallucinations and biases of generative artificial intelligence while 892
respecting the privacy of the public and complying with all applicable 893
state laws and policies. Beginning on July 1, 2025, the commissioner 894
shall make such trainings available to state agency employees not less 895
than annually. 896
Sec. 14. Subsection (b) of section 4-124w of the 2024 supplement to the 897
general statutes is repealed and the following is substituted in lieu 898
thereof (Effective July 1, 2024): 899
(b) The department head of the Office of Workforce Strategy shall be 900
the Chief Workforce Officer, who shall be appointed by the Governor in 901
Committee Bill No. 2
LCO No. 1489 31 of 37
accordance with the provisions of sections 4-5 to 4-8, inclusive, with the 902
powers and duties therein prescribed. The Chief Workforce Officer shall 903
be qualified by training and experience to perform the duties of the 904
office as set forth in this section and shall have knowledge of publicly 905
funded workforce training programs. The Chief Workforce Officer shall: 906
(1) Be the principal advisor for workforce development policy, 907
strategy and coordination to the Governor; 908
(2) Be the lead state official for the development of employment and 909
training strategies and initiatives; 910
(3) Be the chairperson of the Workforce Cabinet, which shall consist 911
of agencies involved with employment and training, as designated by 912
the Governor pursuant to section 31-3m. The Workforce Cabinet shall 913
meet at the direction of the Governor or the Chief Workforce Officer; 914
(4) Be the liaison between the Governor, the Governor's Workforce 915
Council, established pursuant to section 31-3h and any local, regional, 916
state or federal organizations and entities with respect to workforce 917
development policy, strategy and coordination, including, but not 918
limited to, implementation of the Workforce Innovation and 919
Opportunity Act of 2014, P.L. 113-128, as amended from time to time; 920
(5) Develop, and update as necessary, a state workforce strategy in 921
consultation with the Governor's Workforce Council and the Workforce 922
Cabinet and subject to the approval of the Governor. The Chief 923
Workforce Officer shall submit, in accordance with the provisions of 924
section 11-4a, the state workforce strategy to the joint standing 925
committees of the General Assembly having cognizance of matters 926
relating to appropriations, commerce, education, higher education and 927
employment advancement, and labor and public employees at least 928
thirty days before submitting such state workforce strategy to the 929
Governor for his or her approval; 930
(6) Coordinate workforce development activities (A) funded through 931
Committee Bill No. 2
LCO No. 1489 32 of 37
state resources, (B) funded through funds received pursuant to the 932
Workforce Innovation and Opportunity Act of 2014, P.L. 113-128, as 933
amended from time to time, or (C) administered in collaboration with 934
any state agency for the purpose of furthering the goals and outcomes 935
of the state workforce strategy approved by the Governor pursuant to 936
subdivision (5) of this subsection and the workforce development plan 937
developed by the Governor's Workforce Council pursuant to the 938
provisions of section 31-11p; 939
(7) Collaborate with the regional workforce development boards to 940
adapt the best practices for workforce development established by such 941
boards for state-wide implementation, if possible; 942
(8) Coordinate measurement and evaluation of outcomes across 943
education and workforce development programs, in conjunction with 944
state agencies, including, but not limited to, the Labor Department, the 945
Department of Education and the Office of Policy and Management; 946
(9) Notwithstanding any provision of the general statutes, review any 947
state plan for each program set forth in Section 103(b) of the Workforce 948
Innovation and Opportunity Act of 2014, P.L. 113-128, as amended from 949
time to time, before such plan is submitted to the Governor; 950
(10) Establish methods and procedures to ensure the maximum 951
involvement of members of the public, the legislature and local officials 952
in workforce development policy, strategy and coordination; 953
(11) In conjunction with one or more state agencies enter into such 954
contractual agreements, in accordance with established procedures and 955
the approval of the Secretary of the Office of Policy and Management, 956
as may be necessary to carry out the provisions of this section. The Chief 957
Workforce Officer may enter into agreements with other state agencies 958
for the purpose of performing the duties of the Office of Workforce 959
Strategy, including, but not limited to, administrative, human resources, 960
finance and information technology functions; 961
Committee Bill No. 2
LCO No. 1489 33 of 37
(12) Market and communicate the state workforce strategy to ensure 962
maximum engagement with students, trainees, job seekers and 963
businesses while effectively elevating the state's workforce profile 964
nationally; 965
(13) For the purposes of subsection (a) of section 10-21c identify 966
subject areas, courses, curriculum, content and programs that may be 967
offered to students in elementary and high school in order to improve 968
student outcomes and meet the workforce needs of the state; 969
(14) Issue guidance to state agencies, the Governor's Workforce 970
Council and regional workforce development boards in furtherance of 971
the state workforce strategy and the workforce development plan 972
developed by the Governor's Workforce Council pursuant to the 973
provisions of section 31-11p. Such guidance shall be approved by the 974
Secretary of the Office of Policy and Management, allow for a reasonable 975
period for implementation and take effect not less than thirty days from 976
such approval. The Chief Workforce Officer shall consult on the 977
development and implementation of any guidance with the agency, 978
council or board impacted by such guidance; 979
(15) Coordinate, in consultation with the Labor Department and 980
regional workforce development boards to ensure compliance with 981
state and federal laws for the purpose of furthering the service 982
capabilities of programs offered pursuant to the Workforce Innovation 983
and Opportunity Act, P.L. 113-128, as amended from time to time, and 984
the United States Department of Labor's American Job Center system; 985
(16) Coordinate, in consultation with the Department of Social 986
Services, with community action agencies to further the state workforce 987
strategy; [and] 988
(17) In consultation with the regional workforce development boards 989
established under section 31-3k, the Department of Economic and 990
Community Development and other relevant state agencies, incorporate 991
training concerning artificial intelligence, as defined in section 1 of this 992
Committee Bill No. 2
LCO No. 1489 34 of 37
act, into workforce training programs offered in this state; 993
(18) In consultation with the Department of Economic and 994
Community Development, the Connecticut Academy of Science and 995
Engineering and broadband Internet access service providers, as 996
defined in section 16-330a, design an outreach program for the purpose 997
of promoting access to broadband Internet access service, as defined in 998
said section, in underserved communities in this state, and identify a 999
nonprofit organization to implement and lead such outreach program 1000
under the supervision of the Chief Workforce Officer, the Department 1001
of Economic and Community Development and the Connecticut 1002
Academy of Science and Engineering; and 1003
[(17)] (19) Take any other action necessary to carry out the provisions 1004
of this section. 1005
Sec. 15. (NEW) (Effective July 1, 2024) Not later than July 1, 2025, the 1006
Board of Regents for Higher Education shall establish, on behalf of 1007
Charter Oak State College, a "Connecticut Citizens AI Academy" for the 1008
purpose of curating and offering online courses concerning artificial 1009
intelligence and the responsible use of artificial intelligence. The board 1010
shall, in consultation with Charter Oak State College, develop 1011
certificates and badges to be awarded to persons who successfully 1012
complete such courses. As used in this section, "artificial intelligence" 1013
has the same meaning as provided in section 1 of this act. 1014
Sec. 16. (NEW) (Effective July 1, 2024) (a) As used in this section: 1015
(1) "Artificial intelligence" has the same meaning as provided in 1016
section 1 of this act; 1017
(2) "Generative artificial intelligence system" has the same meaning 1018
as provided in section 1 of this act; and 1019
(3) "Prompt engineering" means the process of guiding a generative 1020
artificial intelligence system to generate a desired output. 1021
Committee Bill No. 2
LCO No. 1489 35 of 37
(b) Not later than July 1, 2025, the Board of Regents for Higher 1022
Education shall establish, on behalf of the regional community-technical 1023
colleges, certificate programs in prompt engineering, artificial 1024
intelligence marketing for small businesses and artificial intelligence for 1025
small business operations. 1026
Sec. 17. (Effective July 1, 2024) Not later than December 31, 2024, the 1027
Department of Economic and Community Development shall: 1028
(1) In collaboration with The University of Connecticut and the 1029
Connecticut State Colleges and Universities, develop a plan to offer 1030
high-performance computing services to businesses and researchers in 1031
this state; 1032
(2) In collaboration with The University of Connecticut, establish a 1033
confidential computing cluster for businesses and researchers in this 1034
state; and 1035
(3) In collaboration with industry and academia, conduct a "CT AI 1036
Symposium" to foster collaboration between academia, government and 1037
industry for the purpose of promoting the establishment and growth of 1038
artificial intelligence businesses in this state. 1039
This act shall take effect as follows and shall amend the following
sections:
Section 1 October 1, 2024 New section
Sec. 2 October 1, 2024 New section
Sec. 3 October 1, 2024 New section
Sec. 4 October 1, 2024 New section
Sec. 5 October 1, 2024 New section
Sec. 6 October 1, 2024 New section
Sec. 7 October 1, 2024 New section
Sec. 8 from passage New section
Sec. 9 October 1, 2024 53a-189c(a)
Sec. 10 July 1, 2024 9-600
Sec. 11 July 1, 2024 New section
Sec. 12 from passage New section
Committee Bill No. 2
LCO No. 1489 36 of 37
Sec. 13 July 1, 2024 4a-2e
Sec. 14 July 1, 2024 4-124w(b)
Sec. 15 July 1, 2024 New section
Sec. 16 July 1, 2024 New section
Sec. 17 July 1, 2024 New section
Statement of Purpose:
To: (1) Establish requirements concerning the development,
deployment and use of certain artificial intelligence systems; (2)
establish an Artificial Intelligence Advisory Council; (3) prohibit
dissemination of certain synthetic images; (4) prohibit distribution of,
and agreements to distribute, certain deceptive media concerning
elections; (5) require state agencies to study potential uses of generative
artificial intelligence and propose pilot projects; (6) require the
Commissioner of Administrative Services to provide training
concerning generative artificial intelligence; (7) require the Chief
Workforce Officer to (A) incorporate artificial intelligence training into
workforce training programs, and (B) design a broadband outreach
program; (8) require the Board of Regents for Higher Education to
establish (A) a "Connecticut Citizens AI Academy", and (B) certificate
programs in fields related to artificial intelligence; and (9) require the
Department of Economic and Community Development to (A) develop
a plan to offer high-performance computing services, (B) establish a
confidential computing cluster, and (C) conduct a "CT AI Symposium".
[Proposed deletions are enclosed in brackets. Proposed additions are indicated by underline, except
that when the entire text of a bill or resolution or a section of a bill or resolution is new, it is not
underlined.]
Co-Sponsors: SEN. LOONEY, 11th Dist.; SEN. DUFF, 25th Dist.
SEN. ANWAR, 3rd Dist.; SEN. CABRERA, 17th Dist.
SEN. FLEXER, 29th Dist.; SEN. GASTON, 23rd Dist.
SEN. HARTLEY, 15th Dist.; SEN. HOCHADEL, 13th Dist.
SEN. KUSHNER, 24th Dist.; SEN. LESSER, 9th Dist.
SEN. MAHER, 26th Dist.; SEN. MARONEY, 14th Dist.
SEN. MARX, 20th Dist.; SEN. MCCRORY, 2nd Dist.
SEN. MILLER P., 27th Dist.; SEN. MOORE, 22nd Dist.
SEN. NEEDLEMAN, 33rd Dist.; SEN. OSTEN, 19th Dist.
SEN. RAHMAN, 4th Dist.; SEN. SLAP, 5th Dist.
SEN. WINFIELD, 10th Dist.
Committee Bill No. 2
LCO No. 1489 37 of 37
S.B. 2