OpenAI Hacked by Researchers Using Claude
Hacktron said the attack chain took less than 72 hours and led to access to employee ChatGPT and Codex accounts before OpenAI fixed the flaws.
- Security startup Hacktron disclosed this week that it used Anthropic's Claude Opus 5 to compromise an OpenAI employee's ChatGPT account by exploiting a vulnerability in the company's community forum.
- Researchers discovered a heap buffer overflow in the forum's libheif image-processing software, which they developed into a functional exploit chain using Claude Opus 5 in less than 72 hours, according to Hacktron.
- After gaining forum access, the team exploited an identity flaw to compromise an employee's Codex account, then used it to generate a benign pull request within OpenAI's internal GitHub repository.
- OpenAI patched the community forum, revoked affected sessions, and paid Hacktron a $6,500 bounty, though the company has not published a detailed account of the incident.
- CEO Zayne Zhang of Hacktron noted that "the worlds of AI safety and cybersecurity are converging," highlighting how AI agents with broad credential access create new attack surfaces across internal infrastructure.
223 Articles
223 Articles
How 3 Indian-origin researchers used Claude to breach OpenAI systems in 72 hours
Three Indian cybersecurity researchers from Hacktron used Anthropic's Claude AI to help exploit vulnerabilities that provided them with access to multiple OpenAI employee ChatGPT and Codex accounts.
How 3 Indian researchers used Claude to hack OpenAI
Discover how three Indian-origin researchers at Hacktron AI used Anthropic’s Claude to uncover and exploit security vulnerabilities at OpenAI in under 72 hours, exposing employee accounts and internal code access—and what this means for the future of AI-driven cybersecurity.
The Washington Post
Attention to the UnseenHackers who broke into OpenAI warn the AI industry has a security problem
Security experts said companies developing powerful AI should step up protections against cyberattacks.
3 Indian-origin researchers used Claude to breach OpenAI systems in 72 hours
Three Indian-origin cybersecurity researchers at Hacktron AI used Anthropics Claude models while investigating vulnerabilities that ultimately gave them access to OpenAI employee accounts and a private GitHub environment, according to reports. Harsh Jaiswal, Mohan Pedhapati and Rahul Maini said the work took less than 72 hours and cost under $3,000 in AI tokens. OpenAI later awarded them a $6,500 bounty.
Researchers at an AI security firm utilized Anthropic's Claude to exploit a security vulnerability in OpenAI, successfully infiltrating the internal system within three days. This experiment demonstrated that AI can reduce the difficulty of cyber attacks, and OpenAI immediately fixed the vulnerability. Warnings continue to mount that the threat of large-scale data leaks and system destruction will increase if state-sponsored hackers utilize AI i…
Coverage Details
Bias Distribution
- 41% of the sources are Center
Factuality
To view factuality data please Upgrade to Premium
