Researchers Exploit Claude to Breach OpenAI Employee Accounts

## Hacktron Team Claims Claude‑Powered Breach A group of three independent security researchers from Hacktron announced that they were able to compromise OpenAI employee accounts in less than 72 hours. The breach was carried out using Anthropic's large‑language models Claude Opus 4.8 and 5, as reported by *The Wall Street Journal*.

## Access to Sensitive Codebase The researchers say the compromised credentials gave them entry to OpenAI's private GitHub repository, known as the "Monorepo." This repository is described as containing the company's core algorithmic implementations, which the report refers to as "OpenAI's algorithmic secrets."

## Source Attribution The details of the incident were first published by *The Verge AI* on September 18, 2026, citing the Wall Street Journal investigation. No additional parties or motives were disclosed in the available excerpt.

## Implications for AI Security While the report focuses on the technical steps taken by the Hacktron team, it underscores the potential risks associated with using advanced language models for malicious purposes. Organizations handling sensitive AI assets may need to reassess access controls and monitoring around LLM‑driven workflows.

*Source: The Verge AI (https://www.theverge.com/ai-artificial-intelligence/997444/openai-hack-claude-heif-heist)*