6 Hrs_1777370074079_638186083.jpg&w=64&q=75)
Source:
تكنولوجيا
_1777370155750_318096895.jpg&w=64&q=75){kind=small}
Source: تكنولوجيا
6 Hrsتكنولوجيا
Source: تكنولوجيا
6 HrsReady to play
Ready to play
OpenAI announced the launch of “GPT-6 Cyber,” a model designed to help security teams find vulnerabilities, verify whether they can be exploited, and develop fixes. The company said the model helped identify 81 high-severity vulnerabilities in Firefox, reproduce 1,724 vulnerabilities in the Linux kernel, and develop exploit techniques for 948 of them. It scored 48.5% on a specific exploit test, compared with 32.1% for its predecessor, and avoided targeting services outside the test’s scope in 99.93% of cases, according to the company’s tests. OpenAI classified the model as “critical” for cybersecurity and therefore made it available to a limited group of approved customers, subject to restrictions and monitoring. The company cautioned that these results came from its own tests and were not an independent evaluation.
Notice: This Is an AI-Generated Summary
Comments (0)