A Bill for an Act
Page 1, Line 101Concerning public safety protection from the risks of
Page 1, Line 102artificial intelligence systems.
Bill Summary
(Note: This summary applies to this bill as introduced and does not reflect any amendments that may be subsequently adopted. If this bill passes third reading in the house of introduction, a bill summary that applies to the reengrossed version of this bill will be available at http://leg.colorado.gov.)
The bill prohibits a developer that has trained a foundation artificial intelligence model (foundation model) from preventing a worker from, or retaliating against a worker for, disclosing or threatening to disclose information to the developer, the attorney general, or appropriate state or federal authorities if the worker has reasonable cause to believe the information indicates one of the following:
- The developer is out of compliance with law applicable to public safety or security;
- The developer's activities pose a substantial risk to public safety or security, even if the developer is not out of compliance with any law; or
- The developer has made false or misleading statements concerning public safety or security or concerning the developer's management of risks to public safety or security.
- Reinstatement or rehiring of a worker, with or without back pay;
- The greater of either $10,000 or any lost pay resulting from the violation;
- Punitive damages; and
- Reasonable attorney fees to an aggrieved worker.
A developer must provide notice to all workers working on a foundation model of the workers' rights and responsibilities under the bill.
The bill requires a developer to create and provide an internal process through which a worker may anonymously disclose information to the developer regarding a risk to public safety or security enabled by the developer's foundation model. The developer shall provide a monthly update to the worker who made the disclosure regarding the status of the developer's investigation of the disclosure and the actions taken by the developer in response to the disclosure.
An aggrieved worker may commence a civil action in district court against a developer for a violation of the bill. A court may order relief against a developer that is found to have violated the bill, including:
This Unofficial Version Includes Committee
Amendments Not Yet Adopted on Second Reading
Page 2, Line 1Be it enacted by the General Assembly of the State of Colorado:
Page 2, Line 2SECTION 1. In Colorado Revised Statutes, add article 13.6 to title 8 as follows:
Page 2, Line 3ARTICLE 13.6
Whistleblower Protections for Workers
Page 2, Line 4Working in Artificial Intelligence
Page 2, Line 58-13.6-101. Definitions.As used in this article 13.6, unless the context otherwise requires:
Page 2, Line 6(1) "Artificial intelligence" has the same meaning as
Page 2, Line 7"artificial intelligence system" as set forth in section 6-1-1701 (2).
Page 3, Line 2(2) "Foundation model" means an artificial intelligence model that:
Page 3, Line 3(a) Is trained on broad data;and
(b) Is applicable across a wide range of contexts.
Page 3, Line 5(3) "Large artificial intelligence developer" means an entity that:
Page 3, Line 6(a) Is an employer as defined in section 8-4-101 (6) or an
Page 3, Line 7entity that contracts with five or more independent contractors in the state each year;
Page 3, Line 8(b) Has, within the preceding five years, trained a
Page 3, Line 9foundation model at a computational cost of at least twenty
Page 3, Line 10million dollars as measured by the average cost of an
Page 3, Line 11equivalent amount of cloud computing at the time that training commences; and
Page 3, Line 12(c) Has, within any twelve-month period during the
Page 3, Line 13preceding five years, trained one or more foundation models at
Page 3, Line 14a total computational cost of at least one hundred million
Page 3, Line 15dollars as measured by the average cost of an equivalent amount of cloud computing at the time that training commences.
Page 3, Line 16(4) "Risk to public safety or security" means a risk that a
Page 3, Line 17foundation model could cause or materially enable death, bodily injury, or damage to property.
Page 3, Line 18(5) (a) "Worker" means an individual who performs
Page 3, Line 19services for and under the control and direction of a large
Page 4, Line 1artificial intelligence developer for wages or other remuneration.
Page 4, Line 2(b) "Worker" includes:
(I) An employee as defined in section 8-4-101 (5);
Page 4, Line 3(II) An independent contractor; or
(III) A corporate officer.
Page 4, Line 48-13.6-102. Prohibition against retaliatory action related to a
Page 4, Line 5substantial risk to public safety or security - limitation on worker
Page 4, Line 6contracts. (1) A large artificial intelligence developer shall not
Page 4, Line 7prevent a worker from, or retaliate against a worker for,
Page 4, Line 8disclosing or threatening to disclose information to the large
Page 4, Line 9artificial intelligence developer, the attorney general, or
Page 4, Line 10appropriate state or federal authorities, including through
Page 4, Line 11terms and conditions of employment or seeking to enforce terms
Page 4, Line 12and conditions of employment, if the worker has reasonable cause to believe the information indicates one of the following:
Page 4, Line 13(a) The large artificial intelligence developer is out of compliance with law applicable to public safety or security;
Page 4, Line 14(b) The large artificial intelligence developer's activities
Page 4, Line 15pose a substantial risk to public safety or security, even if the
Page 4, Line 16large artificial intelligence developer is not out of compliance with any law; or
Page 4, Line 17(c) The large artificial intelligence developer has made
Page 4, Line 18false or misleading statements concerning public safety or
Page 4, Line 19security or the large artificial intelligence developer's management of risks to public safety or security.
Page 4, Line 20(2) This section does not apply to a worker who discloses information:
Page 5, Line 1(a) That the worker knows to be false; or
Page 5, Line 2(b) With reckless disregard for the truth or falsity of the information.
Page 5, Line 3(3) (a) A large artificial intelligence developer shall not
Page 5, Line 4require or attempt to require a worker to sign a contract or
Page 5, Line 5other agreement that would limit or prevent the worker from
Page 5, Line 6disclosing information about risks to public safety or security
Page 5, Line 7or to otherwise abide by a workplace policy that would limit or prevent such disclosures.
Page 5, Line 8(b) A contract or agreement that violates subsection
Page 5, Line 9(3)(a) of this section is void and unenforceable as contrary to
Page 5, Line 10public policy. Alarge artificial intelligence developer's
Page 5, Line 11attempt to impose such a contract or agreement is an adverse action in violation of this article 13.6.
Page 5, Line 128-13.6-103. Written notice required. (1) A large artificial
Page 5, Line 13intelligence developer shall provide a clear notice to all
Page 5, Line 14workers working on a foundation model of the workers' rights
Page 5, Line 15and responsibilities under this article 13.6. A large artificial
Page 5, Line 16intelligence developer complies with this section if the large artificial intelligence developer does either of the following:
Page 5, Line 17(a) (I) At all times, displays within all workplaces
Page 5, Line 18maintained by the large artificial intelligence developer a
Page 5, Line 19notice to all workers of their rights and responsibilities under this article 13.6;
Page 5, Line 20(II) Ensures that all new workers receive the notice
Page 5, Line 21described in subsection (1)(a)(I) of this section; and
Page 6, Line 1(III) Ensures that workers who periodically work
Page 6, Line 2remotely receive the notice described in subsection (1)(a)(I) of this section; or
Page 6, Line 3(b) At least once every calendar year, provides written
Page 6, Line 4notice to all workers of their rights and responsibilities under
Page 6, Line 5this article 13.6 and ensures that the notice is received and acknowledged by all workers.
Page 6, Line 68-13.6-104. Internal reporting process. (1) (a) A large
Page 6, Line 7artificial intelligence developer shall provide a reasonable
Page 6, Line 8internal process through which a worker may anonymously
Page 6, Line 9disclose information to the large artificial intelligence
Page 6, Line 10developer if the worker believes in good faith that the
Page 6, Line 11information indicates the conditions described in section 8-13.6-102 (1).
Page 6, Line 12(b) The internal reporting process must include a monthly
Page 6, Line 13update to the worker who made the disclosure regarding the
Page 6, Line 14status of the large artificial intelligence developer's
Page 6, Line 15investigation of the disclosure and the actions taken by the
Page 6, Line 16large artificial intelligence developer in response to the
Page 6, Line 17disclosure. The monthly updates must continue until the matter
Page 6, Line 18is resolved, at which time the large artificial intelligence developer shall provide a final update to the worker.
Page 6, Line 19(2) (a) The large artificial intelligence developer shall
Page 6, Line 20maintain a disclosure or response of the internal reporting
Page 6, Line 21process for a minimum of seven years after the date when the response is created.
Page 6, Line 22(b) At least once per quarter, the large artificial
Page 7, Line 1intelligence developer shall share each disclosure and response
Page 7, Line 2with the officers and directors of the large artificial intelligence developer who do not have a conflict of interest.
Page 7, Line 38-13.6-105. Relief authorized. (1) An aggrieved worker may
Page 7, Line 4commence a civil action in district court against a large
Page 7, Line 5artificial intelligence developer for a violation of this article 13.6.
Page 7, Line 6(2) In a civil action commenced by an aggrieved worker
Page 7, Line 7under this section, the court may order affirmative relief to the
Page 7, Line 8aggrieved worker against a large artificial intelligence
Page 7, Line 9developer that is found to have violated this article 13.6, including:
Page 7, Line 10(a) Reinstatement or rehiring of the worker, with or without back pay;
Page 7, Line 11(b) The greater of either:
(I) Ten thousand dollars; or
Page 7, Line 12(II) Any lost pay resulting from the violation, including
Page 7, Line 13back pay if the aggrieved worker is reinstated or rehired and
Page 7, Line 14front pay if the aggrieved worker is not reinstated or rehired; and
Page 7, Line 15(c) Any other equitable relief the court deems appropriate.
Page 7, Line 16(3) (a) The court may award an aggrieved worker punitive
Page 7, Line 17damages against a large artificial intelligence developer if the
Page 7, Line 18aggrieved worker demonstrates by clear and convincing
Page 7, Line 19evidence that the large artificial intelligence developer
Page 7, Line 20engaged in a discriminatory, adverse, or retaliatory
Page 8, Line 1employment practice with malice or reckless indifference to the rights of the aggrieved worker.
Page 8, Line 2(b) In determining the appropriate level of damages to
Page 8, Line 3award an aggrieved worker, the court shall consider the
Page 8, Line 4gravity of the risk to public safety or security, the size and
Page 8, Line 5assets of the large artificial intelligence developer, and the
Page 8, Line 6egregiousness of the discriminatory, adverse, or retaliatory employment practice.
Page 8, Line 7(4) If an aggrieved worker seeks punitive damages
Page 8, Line 8pursuant to subsection (3) of this section, any party to the civil action may demand a trial by jury.
Page 8, Line 9(5) The court shall award reasonable attorney fees to an
Page 8, Line 10aggrieved worker if the aggrieved worker prevails in an action brought pursuant to this article 13.6.
Page 8, Line 118-13.6-106. No effect on other law.This article 13.6 does not limit protections provided to workers by any other law.
Page 8, Line 128-13.6-107. Trade secrets.This article 13.6 does not permit
Page 8, Line 13a worker to disclose a trade secret, as defined in 18 U.S.C. sec.
Page 8, Line 141839 (3), when such disclosure is prohibited by 18 U.S.C. secs. 1831 to 1839.
Page 8, Line 15SECTION 2. Act subject to petition - effective date -
Page 8, Line 16applicability. (1) This act takes effect at 12:01 a.m. on the day following
Page 8, Line 17the expiration of the ninety-day period after final adjournment of the
Page 8, Line 18general assembly; except that, if a referendum petition is filed pursuant
Page 8, Line 19to section 1 (3) of article V of the state constitution against this act or an
Page 8, Line 20item, section, or part of this act within such period, then the act, item,
Page 8, Line 21section, or part will not take effect unless approved by the people at the
Page 9, Line 1general election to be held in November 2026 and, in such case, will take
Page 9, Line 2effect on the date of the official declaration of the vote thereon by the governor.
Page 9, Line 3(2) This act applies to conduct occurring on or after the applicable effective date of this act.