TechnologyBreaking NewsBreakingLatest News
OpenAI Releases GPT-6 Cyber Model, Claims Reduced Hacking Risks
Sentiment: NeutralHigh Priority
What Happened?
OpenAI has unveiled its latest AI model, GPT-6 Cyber, designed to identify and fix cybersecurity flaws. The company claims it is less likely to hack compared to previous models.
Key points
- GPT-6 Cyber is classified as 'critical' in cybersecurity, outperforming GPT-6 Astra, which is 'high'.
- The model identified 948 critical exploits in Linux in a single run.
- GPT-6 Cyber avoided exploiting out-of-scope services in 99.93% of cases during evaluations.
- It scored 48.5% on ExploitGym for exploiting specified vulnerabilities, compared to 42.4% for GPT-6 Astra.
- The model helped identify 81 previously unknown Firefox vulnerabilities classified as high severity.
Key insights
- GPT-6 Cyber's improved safeguards may reduce the risk of unauthorized access.
- The model's ability to identify vulnerabilities could lead to more secure software.
- OpenAI's focus on cybersecurity reflects growing concerns over AI misuse.
Why it matters
The release of GPT-6 Cyber could enhance cybersecurity measures for organizations.
Entities
OpenAI· OrganizationGPT-6 Cyber· ProductLinux· TechnologyMozilla· OrganizationDaybreak Red· Company
Source
India Today
