OpenAI is coming for Anthropic’s cybersecurity crown.
The AI company announced last Thursday that it is releasing a limited preview of a model called GPT-5.5-Cyber to vetted cybersecurity professionals–in much the same way that Anthropic announced it had released a preview of Mythos to a select group in early April, and launched a public beta of Claude Security for Claude Enterprise customers later that month.
GPT‑5.5‑Cyber has fewer restrictions than GPT-5.5, which has been nicknamed “Spud.”
“It is designed to make the cyber capabilities of GPT‑5.5 more useful for verified defenders working on defensive tasks, while continuing to restrict requests that could enable real-world harm,” OpenAI detailed in a blog post.
OpenAI touts GPT‑5.5 as the company’s “smartest, most intuitive model for both general-purpose knowledge work and cybersecurity tasks,” and is releasing it via tiered access levels.
“This first preview is not expected to outperform GPT‑5.5 across every cyber evaluation. Instead, it supports an iterative deployment process to both accelerate defenders and safely support more specialized authorized workflows that require more permissive behavior, paired with stronger verification, misuse monitoring, approved-use scoping, and partner feedback,” the blog states.
The blog contains commentary from various companies including Intel, SentinelOne, Okta, and cybersecurity firm Snyk. OpenAI noted that both individuals and enterprises can apply for access to GPT-5.5-Cyber by verifying their identity or communicating through an enterprise representative.
Rival Anthropic announced that Claude Security is in public beta for Claude Enterprise customers across the world to use to defensively scan their own code for vulnerabilities; it can also suggest and deploy patches. The tool actually runs on Opus 4.7, not Claude Mythos, Anthropic’s most powerful and heavily guarded model to-date. It has reportedly demonstrated tremendous capabilities in identifying and exploiting software vulnerabilities.
Meanwhile, Mozilla, the nonprofit that operates browser Mozilla Firefox, announced its preview of Claude Mythos has helped it identify and fix a stunning 271 vulnerabilities–just in the initial evaluation.
“As these capabilities reach the hands of more defenders, many other teams are now experiencing the same vertigo we did when the findings first came into focus,” Mozilla chief technology officer Bobby Holley wrote in an April blog post.
Last Thursday, the team published another blog with details and advice on how to leverage emerging technology for cybersecurity. – Inc./TNS
