OpenAI's new cyber model found ways to attack nearly 1000 Linux flaws
AI is making it possible to uncover and test security weaknesses at rapid speed
TL;DR
- OpenAI's new cyber model identified 1,724 security flaws in Linux, with 948 exploitable for administrator access.
- The model uncovered 10 previously unknown exploitation techniques and 81 high-severity flaws in Firefox.
- OpenAI is expanding access to its cyber models through the Daybreak program and creating a new tier for critical infrastructure operators.
- The advancement comes amid concerns about AI being used by hackers to target vital systems.