OpenAI launched two cybersecurity initiatives in under a week, offering vetted teams access to its GPT-5.5-Cyber model and a broader defensive platform called Daybreak. The moves come one month after rival Anthropic's Claude Mythos Preview proved it could find and patch 271 vulnerabilities in Firefox. The Daybreak initiative, announced May 11, is OpenAI's answer to Anthropic's Project Glasswing. Daybreak uses GPT-5.5 for general security work and GPT-5.5 with Trusted Access for Cyber for defensive workflows including secure code review, vulnerability triage, malware analysis, detection engineering and patch validation. A specialized GPT-5.5-Cyber model handles red teaming and penetration testing.
OpenAI is already working with Cloudflare, Cisco, CloudStrike, Palo Alto Networks, Oracle and Akamai on Daybreak. The company first rolled out GPT-5.5-Cyber on May 7 in a limited preview for vetted cybersecurity teams, CNBC reported. The model is trained to be more permissive on security tasks than the standard GPT-5.5, whose safeguards would have blocked certain workflows.
"GPT-5.5-Cyber lets a smaller set of partners study advanced workflows where specialized access behavior may matter," OpenAI said in a blog post.
Anthropic set the bar a month earlier with Claude Mythos Preview, an unreleased model that powers its Project Glasswing cybersecurity initiative. Mozilla revealed in April that Mythos helped it find and patch 271 vulnerabilities in the latest Firefox release. The results caught Washington's attention: Anthropic CEO Dario Amodei met with senior Trump administration officials to discuss Mythos, and Vice President JD Vance and Treasury Secretary Scott Bessent held calls with leading tech CEOs ahead of the model's release. The cybersecurity push is part of a broader OpenAI enterprise offensive. On May 11. The company also announced the OpenAI Deployment Company, a new venture with more than $4 billion in initial investment from TPG, Advent, Bain Capital and Brookfield.
It is acquiring consulting firm Tomoro, bringing roughly 150 AI engineers and deployment specialists from day one. Tomoro, formed in 2023 in alliance with OpenAI, counts Mattel, Red Bull, Tesco and Virgin Atlantic as clients. The urgency is understandable. Anthropic now commands over 40% of the generative AI coding market compared to OpenAI's 21%, according to AI Grid data.
Anthropic's valuation has surpassed $1 trillion with annualized revenue of $30 billion, while OpenAI's valuation sits at $850 billion. Claude Code alone generates $2.5 billion in annualized revenue.
OpenAI's week of product and corporate announcements signals a company trying to close the gap on multiple fronts. Whether GPT-5.5-Cyber and Daybreak can match Mythos's early proof of concept remains the open question, but the partnerships with Cloudflare, Cisco and Palo Alto Networks give Daybreak infrastructure that Glasswing has yet to publicly match.
