Incidents

Incident, incident type, discovery method, incident reporting deadline and recipient.

AI incident reporting is one of the few places frontier AI safety has moved from voluntary framework to binding law: California SB 53 and New York's RAISE Act both set incident-reporting deadlines with named recipients. This track defines incident, incident type, discovery method, and incident reporting deadline and recipient, crosswalked against those statutes plus Anthropic's, OpenAI's, Google DeepMind's, Meta's, and xAI's own incident-disclosure practice.

• Incident reporting deadline and recipient
  Proposedrule

  A proposed rule stating that a reportable incident must be reported to a named recipient within a set deadline, sometimes shorter for imminent harm.

• Discovery method
  Proposedproperty

  A proposed property recording how an incident was first detected: the channel used, the detecting party, and the delay before detection.

• Incident type
  Proposedcontrolled-value

  A proposed controlled classification of incidents by mechanism and severity, letting similar events be compared across different developers.

• Incident
  Proposedrecord-type

  A proposed record of an event where a model, safeguard, or process caused or nearly caused harm, including dates, status, and response taken.