Researchers Exploit AI Assistant and Library Flaw to Access OpenAI Internal Code
Researchers exploited a flaw in an image-decoding library to access OpenAI's internal code, using an employee's AI coding assistant rather than stealing passwords. The initial vulnerability had been fixed months earlier but lacked a security label, delaying widespread patching. OpenAI's bug bounty paid $6,500 for the internal code access but excluded the forum breach that enabled entry. The team bypassed AI safeguards by disguising their exploit as a training exercise, highlighting new security challenges with AI-connected tools.
First-hand measurement across 6 sources
We measured how 6 outlets covered this story. No outlet gave this story a measurable political slant — there is no left–right reading to report. Overall sentiment is neutral (43/100). Lens Score 40/100.
Outlets measured: wion, wion, wion, wion, wion, wion. See how each one headlined and framed the same story in the source comparison below.
AI Analysis
Sentiment was consistent across outlets (35–52/100), indicating broadly factual reporting rather than editorialising.
Coverage timeline
wion broke this story on 19 Sept, 01:38 pm. Other outlets followed.