TechResearchers reportedly used Claude to hack OpenAI employees’ ChatGPT accounts
A user sharing The Register’s report says three Hacktron researchers chained two vulnerabilities and used the account access to reach an internal OpenAI repository.
TLDR
The Register reports that researchers used Claude to hack OpenAI employees’ ChatGPT accounts. A user sharing the report says three Hacktron researchers used Anthropic’s Claude models to develop the exploit, chaining two vulnerabilities to take over multiple employees’ accounts. The user says they then used that access to reach an internal OpenAI repository by opening a harmless pull request.
Combined views
882
1 Source, first seen 23h ago
Researchers reportedly used Claude to hack OpenAI employees’ ChatGPT accounts
A user sharing The Register’s report says three Hacktron researchers chained two vulnerabilities and used the account access to reach an internal OpenAI repository.
TLDR
The Register reports that researchers used Claude to hack OpenAI employees’ ChatGPT accounts. A user sharing the report says three Hacktron researchers used Anthropic’s Claude models to develop the exploit, chaining two vulnerabilities to take over multiple employees’ accounts. The user says they then used that access to reach an internal OpenAI repository by opening a harmless pull request.
