Researchers hacked OpenAI using Claude — all the way to the internal code repository
· Source: original
Claude helped seize OpenAI employee accounts
🔓
A security team deployed Anthropic Claude against OpenAI's systems — and got as far as seizing the company's employee accounts.
Further down the chain — access to OpenAI's internal code repository. Claude searched for vulnerabilities and exploited them immediately: from discovery to breaking into other people's systems.
The discovered vulnerabilities were reported through the responsible disclosure channel.
The case is being presented as confirmation of the real level of LLM autonomy in cyber tasks: vulnerability discovery and exploitation proceed without a human in the loop. An image was attached to the write-up.
The full chain is broken down here — the piece is dated September 18, 2026
🤖 Interested in AI agents and automation?
Prompts for building AI agents and automations — read up on the topic:
🔗 The entire prompt library · "AI agents" category
A ready-made product on the topic: AI Prompts for Productivity — grab it and put it to use right away.