Proposes eight criteria for escalating AI incidents internationally, derived from regulations like the EU AI Act, tested on ten incidents, and identifies three patterns of under-detection in current designs.
The exposure enabled read/write access and potential impersonation of AI agent accounts, some of which held financial instrument permissions
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CY 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Designing escalation criteria for international AI incident response: criteria, triggers, and thresholds
Proposes eight criteria for escalating AI incidents internationally, derived from regulations like the EU AI Act, tested on ten incidents, and identifies three patterns of under-detection in current designs.