Users of Anthropic Claude and OpenAI Codex are reporting account suspensions and service restrictions when using these AI tools for software reverse engineering and decompilation. These security measures affect projects involving the modification of personal hardware.
Reports from the Reverse Engineer Anything community indicate that Claude now blocks requests for code decompilation. Similarly, an OpenAI user reported an account ban after attempting to reverse engineer Samsung television software, with the company citing cybersecurity abuse policies.
The surge in reverse engineering activity is linked to the open-source project Reverse Engineer Anything, which enables users to analyze electronics and replace factory firmware with custom alternatives. This includes updating legacy IoT devices and modifying smart TVs.
OpenAI representative Tarun Gogineni stated that account bans for reverse engineering are unintended consequences of complex security filter configurations. He noted that the company is working to resolve these issues, as the platform does not explicitly prohibit such research.
Security expert Matt Brown suggests that current AI safety filters are overly restrictive, hindering legitimate research. Consequently, some developers are shifting toward local, open-weight AI models to avoid sudden policy enforcement changes and maintain control over their tools.
