OpenAI Warns of "High" Cybersecurity Risks in New Models
New testing reveals advanced models could help develop zero-day exploits, prompting stricter governance.
Escalating Threat Levels
OpenAI published a security advisory warning that its upcoming artificial intelligence models are approaching a "high" risk level for cybersecurity capabilities. The blog post detailed that internal testing shows these models have significantly improved their ability to identify vulnerabilities and potentially develop zero-day exploits.
The Frontier Risk Council
To mitigate these risks, OpenAI announced the formation of a Frontier Risk Council, a body comprising external cybersecurity experts and internal researchers. The council is tasked with auditing new models before deployment and developing "infrastructure hardening" protocols. - Capture-the-Flag Results: The company disclosed that its latest models …
Archive Access
This article is older than 24 hours. Create a free account to access our 7-day archive.