Three hackers breached OpenAI's code

Using Anthropic's Claude, three researchers from Hacktron AI hacked into OpenAI's GitHub repository and accessed source code. OpenAI paid them $6,500 through its bug bounty program.

“We're just three guys with Claude and Codex subscriptions”

They found OpenAI's secret sauce

The hackers discovered Monorepo, OpenAI's proprietary code that makes its models operate faster. This is considered the company's core competitive advantage.

AI models find holes faster than fixes

This breach is part of a pattern where AI models discover cybersecurity vulnerabilities faster than companies can patch them, creating a dangerous security gap.

“AI models are finding cybersecurity holes faster than they can be patched”

Both companies call for safety guardrails

OpenAI and Anthropic have both called for safety guardrails to slow AI development, allowing humans to review security threats before models are released publicly.

Part of a larger pattern of breaches

This hack is the latest in a series of security incidents involving both OpenAI and Anthropic, with both companies disclosing breaches and vulnerabilities.