The Next Web
Three researchers used Claude to reach OpenAI’s internal code. OpenAI paid $6,500 and closed the hole in 14 hours.
Three researchers at Hacktron AI used Anthropic’s Claude to chain two weaknesses and reach OpenAI employee accounts and an internal code repository, in under 72 hours. They disclosed privately, OpenAI patched the single sign-on flaw in…
View original source — The Next Web ↗
Related stories
The Verge
TechnologySep 18, 2026 · 1 min
Security researchers used Claude to help them hack into OpenAI
The Verge
Ars Technica
TechnologySep 18, 2026 · 1 min
Researchers used Claude to hack OpenAI
Ars Technica
TechCrunch
TechnologySep 18, 2026 · 1 min
Researchers used Anthropic’s Claude to hack into OpenAI
TechCrunch
TechRadar
TechnologySep 18, 2026 · 1 min
White hat hackers just breached OpenAI using Anthropic's Claude in less than 72 hours
TechRadar