[Researchers successfully used the Claude model to invade OpenAI and enter the internal code system] "Science and Technology Innovation Board Daily" reported on the 18th that it was reported that the independent information security research team Hacktron AI used Anthropic's Claude model to find and exploit a series of vulnerabilities, successfully entered the ChatGPT account of an OpenAI employee, and gained access to OpenAI's internal Monorepo private code system. Hacktron AI is composed of top hackers and security researchers. It is an independent team invited to participate in the OpenAI vulnerability bounty. This intrusion test is an authorized security research. OpenAI later paid Hacktron AI a $6,500 bounty and said it had solved the security issues discovered by the researchers.