1memoGet 1memo+
Back to memos

2 updates · 1 appSeptember 19, 2026

OpenAI faces security, legal reports

Security researchers reportedly used a rival AI to hack into OpenAI employee accounts, gaining access to internal company code.

ChatGPT

2 updates

Expanded summary · ChatGPT

Security researchers reportedly used a rival AI to hack into OpenAI employee accounts, gaining access to internal company code.

A team of independent security researchers said they used Anthropic's Claude to help them find and exploit security holes at OpenAI. They were able to take over employee accounts and see internal code.

The researchers said the process took less than 72 hours.

Original reportingSecurity researchers used Claude to help them hack into OpenAI (reported by The Verge)Read original

Expanded summary · ChatGPT

Newly revealed court documents show OpenAI's own teams warned that its web data collection could create a 'doom loop' for publishers. (reported by The Verge)

Internal documents from the New York Times' lawsuit against OpenAI and Microsoft have been made public.

They suggest the companies knew that training their AI on huge amounts of web data could harm websites and publishers who create that content.

Original reportingOpenAI and Microsoft knew they were starting a ‘doom loop’ for the web (reported by The Verge)Read original
Additional sourcesTechCrunch