[Bug] Claude Code fabricates user messages and tool results, impersonating user input

Related Claude Code issues

Matched on shared title terms, labels, topic and reported release — ranked so issues with real discussion come first.

• [Bug] Agent bypasses user restrictions by impersonating human user
  Resolved 💬 2 v2.1.80 #36579
• [BUG] Buddy/Companion can inject ghost messages into the input stream as role: "user", impersonating the user
  Resolved 💬 6 v2.1.96 #45279
• [BUG] Model fabricates entire conversation turns (fake user messages + fake tool results) after an interrupted tool call under transmission latency
  Resolved 💬 5 v2.1.185 #70148
• Claude fabricates test results - reports ALL PASSED when tests are FAILING
  Resolved 💬 5 #46940
• Malicious Google Ad impersonating Claude Code — distributes macOS malware
  Resolved 💬 3 #28665
• [Bug] Agent fabricates test results instead of executing tests
  Resolved 💬 3 v2.1.121 #54349
• [Bug] Claude Code fabricates user responses in auto mode without user confirmation
  Open 💬 0 v2.1.241 #90489
• [BUG] Model generates phantom message impersonating user after background task notifications complete
  Resolved 💬 4 v2.1.81 #37602

Browse by topic

Browse all Claude Code issue topics →