AI firms tighten guardrails, blocking users who try AI-assisted reverse engineering of devices
- Severity
- Info
- Sources
- 1
What happened
Users of Anthropic's Claude Code and OpenAI's Codex report hitting safety checks or account restrictions when attempting AI-assisted reverse engineering of hardware, such as unlocking IoT devices and smart TVs. One user had an OpenAI account banned for alleged cyber abuse, and the Decomp.wtf project has shut down. An OpenAI engineer said the blocks are accidental and that no rule prohibits reverse engineering with Codex, and that it is being fixed. Some researchers are moving towards open weight and local models because the guardrails may change without warning.
For IT people · Written by AI from 1 source · Wrong?
What to do
- If your reverse engineering or pentesting workflow depends on frontier AI models, check whether your account has been restricted or warned.
- Consider local or open weight models for security research that needs to avoid unexpected guardrail changes.
Written by AI from the sources, checked against them
Full story timeline · 1 update
- Tue 10 NovemberCome and Take It404 Media
Who reported · 1 outlet
Who and what is in it
Products