EveryAILaw.comCalifornia SB 53 (Frontier AI Transparency Act)
Obligations Covered
Provisions (2)
Frontier AI Framework Publication #
Requirements
| Requirement | Details |
|---|---|
| Publish frontier AI framework | LFDs must write, implement, and clearly publish a frontier AI framework covering governance structures, catastrophic risk mitigation, cybersecurity practices, and standards alignment (§ 22757.12) |
| Annual update | Framework must be updated and made public at least annually, and within 30 days of any material modification |
| Redaction allowance | LFDs may redact trade secrets, cybersecurity practices, or national security items; unredacted versions retained 5 years |
| Incident reporting to OES | Transmit summary of catastrophic risk assessments to California OES (§ 22757.13) |
| Whistleblower channel | Establish anonymous internal channel for covered employees to report safety concerns |
Penalties
| Violation | Fine |
|---|---|
| Per violation | Up to $1M |
Incident Reporting #
Requirements
| Requirement | Details |
|---|---|
| 15-day OES report | Report critical safety incidents to OES within 15 days of discovery (§ 22757.13(c)(1)) |
| 24-hour imminent-risk report | If incident poses imminent risk of death or serious injury, disclose within 24 hours to appropriate authority including law enforcement (§ 22757.13(c)(2)) |
| OES public mechanism | OES must establish public reporting mechanism for critical safety incidents (§ 22757.13(a)) |
| Catastrophic risk summaries | LFDs must confidentially submit catastrophic-risk assessment summaries to OES (§ 22757.13(b)) |
| Whistleblower protection | Protected channels for employees to anonymously report safety concerns to the developer |
Penalties
| Violation | Fine |
|---|---|
| Per violation | Up to $1M |