[Bug] Anthropic API Error: False Positive Cyber Safeguard Triggered on Benign Text "ola"
Status Open
Reported on v2.1.281
Maintainer reply None cached
Activity 0 comments · opened Sep 23, 2026
Bug Description
False positive on [cyber] safeguard — triggered by "ola" (hello in Portuguese)
The safeguard flagged a simple greeting ("ola") as a cyber threat. No exploit, no command, no malicious intent — just a hello message.
Context: I'm using Claude Code on Kali Linux as root, which may have caused the model to apply the cyber filter preemptively based on environment rather than message content.
Both Fable 5.1 and Opus 4.8 blocked the message before switching. This is a clear false positive.
Request ID: req_011CfM12xN1HhViGNobBkrPi
Environment Info
- Platform: linux
- Terminal: xterm-256color
- Version: 2.1.281
- Feedback ID: efe9bfee-f244-4d07-a32f-aa532a5b4a8b
Errors
[]