STATE OF NEW YORK
________________________________________________________________________
6453--A
2025-2026 Regular Sessions
IN ASSEMBLY
March 5, 2025
___________
Introduced by M. of A. BORES, LASHER, SEAWRIGHT, PAULIN, TAPIA, RAGA,
SHIMSKY, REYES, EPSTEIN, BURKE, HEVESI, P. CARROLL, ZACCARO, HYNDMAN,
LUPARDO, KASSAY, LEE, DAVILA, SCHIAVONI, LUNSFORD, K. BROWN, TANNOUS-
IS, TORRES, HOOKS, GIBBS, ROMERO, COLTON, CONRAD, MEEKS, GLICK, CRUZ,
CUNNINGHAM, FORREST, CHANDLER-WATERMAN, STIRPE -- read once and
referred to the Committee on Science and Technology -- committee
discharged, bill amended, ordered reprinted as amended and recommitted
to said committee
AN ACT to amend the general business law, in relation to the training
and use of artificial intelligence frontier models
The People of the State of New York, represented in Senate and Assem-bly, do enact as follows:
1 Section 1. Short title. This act shall be known and may be cited as
2 the "Responsible AI safety and education act" or "RAISE act".
3 § 2. The general business law is amended by adding a new article 44-B
4 to read as follows:
5 ARTICLE 44-B
6 RESPONSIBLE AI SAFETY AND EDUCATION (RAISE) ACT
7 Section 1420. Definitions.
8 1421. Transparency requirements regarding frontier model train-
9 ing and use.
10 1422. Protections, rights and obligations of employees.
11 1423. Violations.
12 1424. Duties and obligations.
13 1425. Scope.
14 1426. Severability.
15 § 1420. Definitions. As used in this article, the following terms
16 shall have the following meanings:
17 1. "Appropriate redactions" means redactions to a safety and security
18 protocol or audit report that a developer may make when necessary to:
EXPLANATION--Matter in italics (underscored) is new; matter in brackets
[] is old law to be omitted.
LBD00047-12-5
A. 6453--A 2
1 (a) protect public safety to the extent the developer can reasonably
2 predict such risks;
3 (b) protect trade secrets;
4 (c) prevent the release of confidential information as required by
5 state or federal law;
6 (d) protect employee or customer privacy; or
7 (e) prevent the release of information otherwise controlled by state
8 or federal law.
9 2. "Artificial intelligence" means a machine-based system that can,
10 for a given set of human-defined objectives, make predictions, recommen-
11 dations, or decisions influencing real or virtual environments, and that
12 uses machine- and human-based inputs to perceive real and virtual envi-
13 ronments, abstract such perceptions into models through analysis in an
14 automated manner, and use model inference to formulate options for
15 information or action.
16 3. "Artificial intelligence model" means an information system or
17 component of an information system that implements artificial intelli-
18 gence technology and uses computational, statistical, or machine-learn-
19 ing techniques to produce outputs from a given set of inputs.
20 4. "Compute cost" means the cost incurred to pay for compute used in
21 training a model when calculated using the average published market
22 prices of cloud compute in the United States at the start of training
23 such model as reasonably assessed by the person doing the training.
24 5. "Deploy" means to use a frontier model or to make a frontier model
25 foreseeably available to one or more third parties for use, modifica-
26 tion, copying, or a combination thereof with other software, except for
27 training or developing the frontier model, evaluating the frontier model
28 or other frontier models, or complying with federal or state laws.
29 6. "Frontier model" means either of the following:
30 (a) an artificial intelligence model trained using greater than 10§26
31 computational operations (e.g., integer or floating-point operations),
32 the compute cost of which exceeds one hundred million dollars; or
33 (b) an artificial intelligence model produced by applying knowledge
34 distillation to a frontier model as defined in paragraph (a) of this
35 subdivision.
36 7. "Critical harm" means the death or serious injury of one hundred or
37 more people or at least one billion dollars of damages to rights in
38 money or property caused or materially enabled by a large developer's
39 creation, use, storage, or release of a frontier model, through either
40 of the following:
41 (a) The creation or use of a chemical, biological, radiological, or
42 nuclear weapon; or
43 (b) An artificial intelligence model engaging in conduct that does
44 both of the following:
45 (i) Acts with limited human intervention; and
46 (ii) Would, if committed by a human, constitute a crime specified in
47 the penal law that requires intent, recklessness, or gross negligence,
48 or the solicitation or aiding and abetting of such a crime.
49 A harm inflicted by an intervening human actor shall not be deemed to
50 result from a developer's activities unless such activities made it
51 substantially easier or more likely for the actor to inflict such harm.
52 8. "Knowledge distillation" means any supervised learning technique
53 that uses a larger artificial intelligence model or the output of a
54 larger artificial intelligence model to train a smaller artificial
55 intelligence model with similar or equivalent capabilities as the larger
56 artificial intelligence model.
A. 6453--A 3
1 9. "Large developer" means a person that has trained at least one
2 frontier model, the compute cost of which exceeds five million dollars,
3 and has spent over one hundred million dollars in compute costs in
4 aggregate in training frontier models. Accredited colleges and univer-
5 sities shall not be considered large developers under this article to
6 the extent that such colleges and universities are engaging in academic
7 research. If a person subsequently transfers full intellectual property
8 rights of the frontier model to another person (including the right to
9 resell the model) and retains none of those rights for themself, then
10 the receiving person shall be considered the large developer and shall
11 be subject to the responsibilities and requirements of this article
12 after such transfer.
13 10. "Model weight" means a numerical parameter in an artificial intel-
14 ligence model that is adjusted through training and that helps determine
15 how inputs are transformed into outputs.
16 11. "Person" means an individual, proprietorship, firm, partnership,
17 joint venture, syndicate, business trust, company, corporation, limited
18 liability company, association, committee, or any other nongovernmental
19 organization or group of persons acting in concert.
20 12. "Safety and security protocol" means documented technical and
21 organizational protocols that:
22 (a) Specify reasonable protections and procedures that, if successful-
23 ly implemented would appropriately reduce the risk of critical harm;
24 (b) Describe reasonable administrative, technical, and physical
25 cybersecurity protections for frontier models within the large develop-
26 er's control that, if successfully implemented, appropriately reduce the
27 risk of unauthorized access to, or misuse of, the frontier models lead-
28 ing to critical harm, including by sophisticated actors;
29 (c) Describe in detail the testing procedure to evaluate if the fron-
30 tier model poses an unreasonable risk of critical harm;
31 (d) Describe in detail how the testing procedure assesses whether the
32 frontier model could be misused, be modified, be executed with increased
33 computational resources, evade the control of its large developer or
34 user, be combined with other software or be used to create another fron-
35 tier model in a manner that would increase the risk of critical harm;
36 (e) State compliance requirements with sufficient detail and specif-
37 icity to allow the large developer or a third party to readily ascertain
38 whether the requirements of the safety and security protocol have been
39 followed;
40 (f) Describe how the large developer will fulfill their obligations
41 under this article, including with respect to any requirements, safe-
42 guards, or modifications; and
43 (g) Designate senior personnel to be responsible for ensuring compli-
44 ance.
45 13. "Safety incident" means an incident of the following kinds that
46 occurs in such a way that it provides demonstrable evidence of an
47 increased risk of critical harm:
48 (a) A frontier model autonomously engaging in behavior other than at
49 the request of a user;
50 (b) Theft, misappropriation, malicious use, inadvertent release, unau-
51 thorized access, or escape of the model weights of a frontier model;
52 (c) The critical failure of any technical or administrative controls,
53 including controls limiting the ability to modify a frontier model; or
54 (d) Unauthorized use of a frontier model.
55 14. "Trade secret" means any form and type of financial, business,
56 scientific, technical, economic, or engineering information, including a
A. 6453--A 4
1 pattern, plan, compilation, program device, formula, design, prototype,
2 method, technique, process, procedure, program, or code, whether tangi-
3 ble or intangible, and whether or how stored, compiled, or memorialized
4 physically, electronically, graphically, photographically or in writing,
5 that:
6 (a) Derives independent economic value, actual or potential, from not
7 being generally known to, and not being readily ascertainable by proper
8 means by, other persons who can obtain economic value from its disclo-
9 sure or use; and
10 (b) Is the subject of efforts that are reasonable under the circum-
11 stances to maintain its secrecy.
12 § 1421. Transparency requirements regarding frontier model training
13 and use. 1. Before deploying a frontier model, the large developer of
14 such frontier model shall do all of the following:
15 (a) Implement a written safety and security protocol;
16 (b) Retain an unredacted copy of the safety and security protocol,
17 including records and dates of any updates or revisions. Such unredacted
18 copy of the safety and security protocol, including records and dates of
19 any updates or revisions, shall be retained for as long as a frontier
20 model is deployed plus five years;
21 (c) (i) Conspicuously publish a copy of the safety and security proto-
22 col with appropriate redactions and transmit a copy of such redacted
23 safety and security protocol to the division of homeland security and
24 emergency services;
25 (ii) Grant the division of homeland security and emergency services or
26 the attorney general access to the safety and security protocol, with
27 redactions only to the extent required by federal law, upon request;
28 (d) Record, as and when reasonably possible, and retain for as long as
29 the frontier model is deployed plus five years information on the
30 specific tests and test results used in any assessment of the frontier
31 model that provides sufficient detail for third parties to replicate the
32 testing procedure; and
33 (e) Implement appropriate safeguards to prevent unreasonable risk of
34 critical harm.
35 2. A large developer shall not deploy a frontier model if doing so
36 would create an unreasonable risk of critical harm.
37 3. A large developer shall conduct an annual review of any safety and
38 security protocol required by this section to account for any changes
39 to the capabilities of their frontier models and industry best practices
40 and, if necessary, make modifications to such safety and security proto-
41 col. If any modifications are made, the large developer shall publish
42 the safety and security protocol in the same manner as required pursuant
43 to paragraph (c) of subdivision one of this section.
44 4. (a) Beginning on the effective date of this article, or ninety days
45 after a developer first qualifies as a large developer, whichever is
46 later, a large developer shall annually retain a third party to perform
47 an independent audit of compliance with the requirements of this
48 section. Such third party shall conduct audits consistent with best
49 practices.
50 (b) The third party shall be granted access to unredacted materials as
51 necessary to comply with the third party's obligations under this subdi-
52 vision.
53 (c) The third party shall produce a report including all of the
54 following:
55 (i) A detailed assessment of the large developer's steps to comply
56 with the requirements of this section;
A. 6453--A 5
1 (ii) If applicable, any identified instances of noncompliance with the
2 requirements of this section, and any recommendations for how the devel-
3 oper can improve its policies and processes for ensuring compliance with
4 the requirements of this section;
5 (iii) A detailed assessment of the large developer's internal
6 controls, including its designation and empowerment of senior personnel
7 responsible for ensuring compliance by the large developer, its employ-
8 ees, and its contractors; and
9 (iv) The signature of the lead auditor certifying the results of the
10 audit.
11 (d) The large developer shall retain an unredacted copy of the report
12 for as long as a frontier model is deployed plus five years.
13 (e) (i) The large developer shall conspicuously publish a copy of the
14 third party's report with appropriate redactions and transmit a copy of
15 such redacted report to the division of homeland security and emergency
16 services.
17 (ii) The large developer shall grant the division of homeland security
18 and emergency services or the attorney general access to the third
19 party's report, with redactions only to the extent required by federal
20 law, upon request.
21 5. A large developer shall disclose each safety incident affecting
22 the frontier model to the division of homeland security and emergency
23 services within seventy-two hours of the large developer learning of the
24 safety incident or within seventy-two hours of the large developer
25 learning facts sufficient to establish a reasonable belief that a safety
26 incident has occurred. Such disclosure shall include: (a) the date of
27 the safety incident; (b) the reasons the incident qualifies as a safety
28 incident as defined in subdivision thirteen of section fourteen hundred
29 twenty of this article; and (c) a short and plain statement describing
30 the safety incident.
31 6. A large developer shall not knowingly make false or materially
32 misleading statements or omissions in or regarding documents produced
33 pursuant to this section.
34 7. Any person who is not a large developer, but who sets out to train
35 a frontier model that if completed as planned would qualify such person
36 as a large developer (i.e. at the end of the training, such person will
37 have spent five million dollars in compute costs on one frontier model
38 and one hundred million dollars in compute costs in aggregate in train-
39 ing frontier models, excluding accredited colleges and universities to
40 the extent such colleges and universities are engaging in academic
41 research) shall, before training such model:
42 (a) Implement a written safety and security protocol, excluding the
43 requirements described in paragraphs (c) and (d) of subdivision twelve
44 of section fourteen hundred twenty of this article; and
45 (b) Transmit a copy of an appropriately redacted safety and security
46 protocol to the division of homeland security and emergency services.
47 § 1422. Protections, rights and obligations of employees. 1. A large
48 developer or a contractor or subcontractor of a large developer shall
49 not prevent an employee from disclosing, or threatening to disclose, or
50 retaliate against an employee for disclosing or threatening to disclose,
51 information to the large developer or the attorney general, if the
52 employee has reasonable cause to believe that the large developer's
53 activities pose an unreasonable or substantial risk of critical harm,
54 regardless of the employer's compliance with applicable law.
55 2. An employee harmed by a violation of this section may petition a
56 court for appropriate temporary or preliminary injunctive relief.
A. 6453--A 6
1 3. A large developer shall inform employees of their protections,
2 rights and obligations under this article within ninety days of the
3 effective date of this article or of becoming a large developer, which-
4 ever is later, upon commencement of employment, and by posting a notice
5 thereof. Such notice shall be posted conspicuously in easily accessible
6 and well-lighted places customarily frequented by employees.
7 4. Nothing in this section shall be deemed to diminish the rights,
8 privileges, or remedies of any employee under any other law or regu-
9 lation or under any collective bargaining agreement or employment
10 contract.
11 5. As used in this section, the following terms shall have the follow-
12 ing meanings:
13 (a) "Employee" has the same meaning as defined in subdivision five of
14 section two of the labor law and includes both of the following:
15 (i) Contractors or subcontractors and unpaid advisors involved with
16 assessing, managing, or addressing the risk of critical harm from fron-
17 tier models; and
18 (ii) Corporate officers.
19 (b) "Contractor or subcontractor" means any person, sole proprietor,
20 partnership, firm, corporation, limited liability company, association
21 or other legal entity who by oneself or through others offers to under-
22 take, or holds oneself out as being able to undertake, or does undertake
23 work assessing, managing, or addressing the risk of critical harm from
24 frontier models on behalf of the large developer.
25 § 1423. Violations. 1. The attorney general may bring a civil action
26 for a violation of this article and to recover all of the following:
27 (a) For a violation of section fourteen hundred twenty-one of this
28 article, a civil penalty in an amount not exceeding ten million dollars
29 for a first violation and in an amount not exceeding thirty million
30 dollars for any subsequent violation.
31 (b) For a violation of section fourteen hundred twenty-two of this
32 article, a civil penalty in an amount not exceeding ten thousand dollars
33 per employee for each violation of such section to be awarded to the
34 employee who was retaliated against.
35 (c) For a violation of section fourteen hundred twenty-one or fourteen
36 hundred twenty-two of this article, injunctive or declaratory relief.
37 2. (a) A provision within a contract or agreement that seeks to waive,
38 preclude, or burden the enforcement of a liability arising from a
39 violation of this article, or to shift that liability to any person or
40 entity in exchange for their use or access of, or right to use or
41 access, a large developer's products or services, including by means of
42 a contract of adhesion, is void as a matter of public policy.
43 (b) A court shall disregard corporate formalities and impose joint and
44 several liability on affiliated entities for purposes of effectuating
45 the intent of this section to the maximum extent allowed by law if the
46 court concludes that both of the following are true:
47 (i) The affiliated entities, in the development of the corporate
48 structure among the affiliated entities, took steps to purposely and
49 unreasonably limit or avoid liability; and
50 (ii) As the result of the steps described in subparagraph (i) of this
51 paragraph, the corporate structure of the large developer or affiliated
52 entities would frustrate recovery of penalties, damages, or injunctive
53 relief under this section.
54 3. The division of homeland security and emergency services shall
55 make any critical safety incident disclosure available to the attorney
56 general upon request.
A. 6453--A 7
1 4. This section does not limit the application of other laws.
2 § 1424. Duties and obligations. The duties and obligations imposed by
3 this article are cumulative with any other duties or obligations imposed
4 under other law and shall not be construed to relieve any party from any
5 duties or obligations imposed under other law and do not limit any
6 rights or remedies under existing law.
7 § 1425. Scope. This article shall only apply to frontier models that
8 are developed, deployed, or operating in whole or in part in New York
9 state.
10 § 1426. Severability. If any clause, sentence, paragraph, subdivision,
11 section or part of this article shall be adjudged by any court of compe-
12 tent jurisdiction to be invalid, such judgment shall not affect, impair,
13 or invalidate the remainder thereof, but shall be confined in its opera-
14 tion to the clause, sentence, paragraph, subdivision, section, or part
15 thereof directly involved in the controversy in which such judgment
16 shall have been made.
17 § 3. This act shall take effect on the ninetieth day after it shall
18 have become a law.
