Français

Organismes ayant un pouvoir

California Attorney General — Catastrophic Risks in Artificial Intelligence Foundation Models (SB 53 / TFAIA enforcement)

Type
Government body
Lieu
California, États-Unis — California, USA — but binding on frontier developers doing business in California, so de facto national
Dernière vérification
2026-09-22
Prochaine vérification
2027-03-21

Pas encore traduit — affiché en anglais.

Comment les joindre

  • SB 53 covered-employee reports to the Attorney General · Complaint route
    Vérifié · 2026-09-22

    Complete the online contact form and select the category 'Catastrophic risks in AI foundation models – employee reports (SB 53)'. For employees and contractors of frontier developers reporting catastrophic risk or non-compliance.

    VERIFIED — quoted from https://oag.ca.gov/sb53. This dedicated category on a general form is easy to miss.

  • Dedicated SB 53 email address · Email address
    none published
    Vérification expirée · 2026-09-22

    Ce canal est conservé uniquement comme référence. Sa vérification est absente, date de plus de 180 jours, ou la période de contact n'est pas ouverte actuellement ; ne vous y fiez pas avant une nouvelle vérification.

    n/a

    VERIFIED ABSENCE — the AG's SB 53 page states submissions must go through the online form; no email is given.

  • Homepage · Homepage
    Non contrôlé

Ce qu'il fait

Enforces the Transparency in Frontier Artificial Intelligence Act (SB 53, signed 29 September 2025). Large frontier developers must publish a Frontier AI Framework covering catastrophic-risk identification, mitigation and governance; all frontier developers must publish transparency reports before deployment; critical safety incidents must be reported to Cal OES within 15 days (24 hours if imminent danger). The AG can seek civil penalties of up to $1,000,000 per violation. SB 53 also creates whistleblower protection for covered employees and requires anonymous internal reporting channels. The AG must publish an annual anonymised, aggregated report on employee reports received.

Ses pouvoirs

fine

Évaluation franche

This is, as of now, the ONLY government channel anywhere in the US built specifically to receive reports about CATASTROPHIC risk from frontier models, with statutory whistleblower protection behind it and a $1m-per-violation penalty. It is a real lever and it is badly publicised — the route is a category buried in the AG's generic contact form. Covered employees are the intended users, but the form itself is open. The AG's annual aggregate report means submissions become part of a public record. If you have insider information about a frontier developer's safety practices, this is the highest-value single address in this directory.

Notes

California's Department of Technology may recommend updates to statutory definitions such as the 'frontier model' compute threshold, so the covered set will shift over time.

Sources

Une erreur ?