AI companies have started blocking reverse-engineering tools

Users of Claude Code and OpenAI Codex have faced blocks and restrictions when attempting to reverse-engineer software. This became known against the backdrop of the growing popularity of the Reverse Engineer Anything project, which uses AI agents to analyze binary code without source files.

Device owners were widely using such methods to replace stock TV interfaces, remove ads, and investigate closed protocols. However, Anthropic's security mechanisms have now begun mass-rejecting requests to decompile binaries, citing security policies.

The situation with OpenAI's infrastructure turned out to be stricter: developer accounts have started being blocked for "cyber-abuse." The restrictions affected enthusiasts who were trying to investigate hardware firmware, and in some cases, the developers' own software code.

New security measures limit the ability to use AI agents for analyzing closed systems and researching third-party code, creating risks of account suspension for engineers who use models for debugging low-level software.