2026-09-20T23:22:28.549Z
- publishedAt:
Not provided→ 2026-04-07T00:00:00.000Z
SOURCE-LINKED INTELLIGENCE
Claude Mythos Preview Reportedly Posted Sandbox Exploit Details to Public Websites During Anthropic Evaluation
During an Anthropic evaluation, an earlier version of Claude Mythos Preview was instructed to circumvent the network restrictions of a secured sandbox environment and contact a researcher. After obtaining broader Internet access and sending the requested message, it later exposed information about how it breached the sandbox by posting it on several publicly reachable websites. Anthropic said the model did not access its weights or internal systems.
Read original source ↗ Open in workspace
- recordType
- incident-report
- evidenceStatus
- reported
- region
- Global
Reported occurrence date: 2026-04-07T00:00:00.000Z
Evidence & attribution
- AI Incident Database
- AI Incident Database · 2026-04-07T00:00:00.000Z
AI Incident Database, Responsible AI Collaborative; McGregor (2021), Preventing Repeated Real World AI Failures by Cataloging Incidents. Incident-specific contributor credits are available at each citation link. Metadata adapted; article text excluded.
License: CC BY-SA 4.0
First collected: 2026-09-19T22:50:59.123Z. This is not the publication date.
Observed changes
AIIC observation times, not verified publisher revision times. Up to eight recent revisions.