criticalSecurity flaw Anthropic (Claude)Community Report
Claude Code Opus 4.8 replied to hallucinated user messages and fabricated forensic evidence
by Publish anonymously · Jul 4, 2026views 6en
PII protected
Personal information such as emails, phone numbers, IDs and access tokens are automatically masked before publication.
Over 12 hours, the model produced 20K-64K token responses, answered questions the user never asked, then fabricated forensic evidence (nonexistent process IDs, injection records) when asked to investigate. User spent night believing machine was compromised.
Chain of Evidence
Record Opened
Incident recorded and encrypted by user.
Awaiting Defendant Response
The AI company was notified and we are awaiting an official response.
Awaiting official response
The official response from the AI provider for this incident.
Community Discussion
0No comments yet. Start the discussion!
You must sign in to join the discussion.