Incident

NIKOLAI editorial proposal (unsourced, not attributed to any lab or regulator): an Incident is a record of an event, identified by a persistent identifier, in which a model, safeguard, or organizational process produced or nearly produced a harm or a safety-relevant failure. The record carries at minimum: start/end dates, a scope description, a current status, the model or system involved, a description of its involvement, and any corrective or response measures taken. NIKOLAI does not itself define what threshold of severity makes an event reportable -- that varies by source (see crosswalk) -- only the record shape needed to compare across sources.

This is CASRAI's own proposed definition, not a definition any named organisation has agreed to. See what NIKOLAI is and is not.

Source of record

Where this definition comes from

EU GPAI Code of Practice, Safety and Security Chapter, Commitment 9 / Measure 9.2
“Signatories track, document and report "relevant information about serious incidents along the entire model lifecycle and possible corrective measures". Measure 9.2 lists nine required fields (start/end dates; "the resulting harm and the victim or affected group"; chain of events; "the model involved"; description of the model's involvement; response taken; recommendation to authorities; "a root cause analysis ..., including ... any failures or circumventions of systemic risk mitigations"; related post-market-monitoring patterns).”
https://ec.europa.eu/newsroom/dae/redirection/document/118119
California SB 53, 22757.11(d)
“"Critical safety incident" (22757.11(d); four types, see F2).”
https://leginfo.legislature.ca.gov/faces/billTextClient.xhtml?bill_id=202520260SB53
Anthropic Advanced AI Framework, p.7 (see F4+ for full verbatim text)
“AAF "Critical Safety Incident" (four limbs, below).”
https://www-cdn.anthropic.com/files/4zrzovbb/website/0a58d567024a8b448ff15158ebc3625328dfcc1f.pdf

Crosswalk

How named organisations use this concept

Every row below is a shadow mapping. It is CASRAI's own reading of a published document. No lab, evaluator or regulator named here has declared, endorsed, or been consulted on this mapping. That will change only when an organisation files its own Mapping Declaration — see the non-endorsement policy.

Organisation	Their term, as published	Match	Source
Anthropic Anthropic Advanced AI Framework; Anthropic Risk Report (August 2026)	“AAF "Critical Safety Incident" (four limbs, below). Risk Report incident register in §4.5.8.2 (aiming to "paint a reasonably comprehensive picture") and "Safety process failures: 'historical cases where Anthropic's safety and security posture fell short of our ideal in some way'" (§5.2.1).”	close confidence: medium	Anthropic Advanced AI Framework
OpenAI OpenAI Frontier Governance Framework	“"AI Safety Incident Response Plan (AIRP)"; the FGF uses "AI safety incident / potential AI safety incident / critical safety incident / serious incident" but "does not define any of them".”	none confidence: medium	OpenAI Frontier Governance Framework
Google DeepMind Google DeepMind Frontier Safety Framework v3.1	“"incidents relating to our frontier safety risk domains" (s.1.3.2); no definition.”	none confidence: medium	Google DeepMind Frontier Safety Framework v3.1
xAI xAI Frontier AI Framework, 30 June 2026	“"serious incident" (trigger point) and "Serious AI safety incidents: used, not defined" (s.2, s.3).” The FAIF26 PDF's own metadata /Title reads "Privileged/Confidential DRAFT working FRAMEWORK DOC"; no xAI statement disambiguating draft vs. final status was found (open-VERIFY register item 4). Treat this row's provenance as a draft document, not a confirmed-final xAI policy, until that is resolved.	none confidence: medium	xAI Frontier AI Framework, 30 June 2026
Meta Meta Advanced AI Scaling Framework v2	“"major incident" and "reporting critical incidents as appropriate" (§2.2.2, §2.3.2); undefined.”	none confidence: medium	Meta Advanced AI Scaling Framework v2
EU EU GPAI Code of Practice, Safety and Security Chapter	“"serious incident" is not given a one-sentence definition in the chapter, but Commitment 9 fully operationalises it: Signatories track, document and report "relevant information about serious incidents along the entire model lifecycle and possible corrective measures". Measure 9.2 lists nine required fields (start/end dates; "the resulting harm and the victim or affected group"; chain of events; "the model involved"; description of the model's involvement; response taken; recommendation to authorities; "a root cause analysis ..., including ... any failures or circumventions of systemic risk mitigations"; related post-market-monitoring patterns).”	exact confidence: high	EU GPAI Code of Practice, Safety and Security Chapter
California SB 53 California SB 53	“"Critical safety incident" (22757.11(d); four types, see F2).”	exact confidence: high	California SB 53
UK AI Safety Institute (AISI) Discovery sweep (evaluator ecosystem) -- UK AISI Incident Report, 4 Aug 2026	“UK AISI "Incident Report: unsanctioned agent behaviour during cyber testing" (4 Aug 2026), described as the first public incident report by a government evaluator (discovery) [UV].” unverified: true. Source document cites this row to tag {DEU}, which is not resolved in the NIKOLAI sources table (only {DEV}, "Discovery sweep", is defined there) -- likely the same discovery-sweep source under an inconsistent tag, but no URL is asserted here to avoid fabricating one. The open-VERIFY register (item 7) separately confirms the UK AISI incident report's structure and field names were not read in full this pass.	none confidence: low	Discovery sweep (evaluator ecosystem)
Frontier Model Forum Frontier Model Forum -- Information Sharing, Incident Reporting and Incident Response Issue Brief	“"Incident: not defined; 'The definition and appropriate scope of what is considered an "incident" depends on regulatory context, risk categories, and the capabilities of the systems involved.'"”	none confidence: medium	Frontier Model Forum -- Information Sharing Issue Brief

Related, not mapped

Pointers that are not crosswalk claims

These sources mention this concept but do not define or map it clearly enough to count as a crosswalk row — noted here so the research is visible without overstating it as a mapping.

METR
Independent incident investigation scoped to "seven specific questions" (RL -- a related pointer, not an incident-record-shape mapping).
METR -- OpenAI/Hugging Face Incident Investigation

Divergence

Where sources materially disagree

The false-friends register (ALIGNMENT-MATRIX.md section 4, row "incident") flags 'incident' itself as a label collision: SB 53 and the EU GPAI chapter give the word an operational legal definition (four limbs / nine required fields), while OpenAI's FGF, Google DeepMind's FSF, xAI's FAIF26, Meta's framework, and even the Frontier Model Forum's own issue brief use the word without defining it -- FMF explicitly declines to fix a scope, stating that the appropriate scope of 'incident' depends on regulatory context, risk categories, and system capabilities. The EQ/DU score split in this crosswalk (EQ for SB53 and the EU chapter, DU for every lab and for FMF) reflects that same real disagreement in the sources, not a scoring inconsistency.