OpenAI says its flagship GPT-5.6 Sol model has set a new state of the art in cybersecurity, solving the UK AI Security Institute's demanding "The Last Ones" cyber range in seven of ten attempts and outpacing both an earlier OpenAI model and a leading rival system. The company framed the result not as an offensive milestone but as evidence that the same capability is beginning to translate into defensive outcomes, helping engineering teams find, validate and fix vulnerabilities in real-world code through its Codex Security product.
Continue reading
The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in✓ Signed in — this article isn’t included in your current plan.