# Researchers used Claude to breach OpenAI’s internal systems

https://mins.news/en/item/65146

- CGTN, 2026-09-19 02:30 UTC
- Tech
- Original: https://news.cgtn.com/news/2026-09-19/Researchers-used-Claude-to-breach-OpenAI-s-internal-systems-1QyI5HU07iE/p.html?UTM_Source=cgtn&UTM_Medium=rss&UTM_Campaign=World

> Security firm Hacktron says it used Anthropic’s Claude to identify and exploit a flaw in OpenAI’s public help forum, gaining control of the site. OpenAI patched the issue in about 14 hours and paid a $6,500 reward. Researchers said Claude Opus 5 produced a working hack in roughly three hours, highlighting concerns that AI is lowering the cost and time needed for sophisticated cyberattacks.

## Stories

- [AI Cybersecurity and Loss of Control Risks](https://mins.news/en/topic/ai-cybersecurity-and-control) — Markdown: https://mins.news/en/topic/ai-cybersecurity-and-control.md

## Related coverage

- [Anthropic’s Claude helped researchers breach OpenAI in under 72 hours](https://mins.news/en/item/65060) — CryptoSlate, 2026-09-19
- [OpenAI hacked “ethically” with Anthropic’s Claude chatbot](https://mins.news/en/video/64967) — Global News, 2026-09-18
- [Security researchers used Claude to help hack into OpenAI](https://mins.news/en/item/64492) — The Verge, 2026-09-18