Key Takeaways
- GPT-5.4-Cyber has been deployed to a limited group of authenticated cybersecurity experts
- The debut follows Anthropic’s Mythos security model by seven days
- Federal Reserve Chair Jerome Powell and Treasury Secretary Scott Bessent expressed alarm over Mythos capabilities
- User access will scale from hundreds to thousands of validated security professionals
- Ramp payment data reveals approximately 33% of American companies subscribed to Anthropic services in March
OpenAI has unveiled a specialized artificial intelligence system designed exclusively for cybersecurity applications. Dubbed GPT-5.4-Cyber, the platform is currently accessible only to authenticated professionals.
We’re expanding Trusted Access for Cyber with additional tiers for authenticated cybersecurity defenders.
Customers in the highest tiers can request access to GPT-5.4-Cyber, a version of GPT-5.4 fine-tuned for cybersecurity use cases, enabling more advanced defensive workflows.…
— OpenAI (@OpenAI) April 14, 2026
The announcement arrived on April 15, Tuesday. This rollout occurred merely seven days following Anthropic’s restricted deployment of Mythos, its competing security-oriented model.
According to OpenAI, GPT-5.4-Cyber represents a “more permissive” variant of the standard GPT-5.4 system. The modification reduces restriction thresholds for cybersecurity operations that the company classifies as authorized defensive activities.
Current availability remains confined to members of OpenAI’s Trusted Access for Cyber initiative. This program, which debuted in February, mandates identity authentication for all users seeking model access.
The company announced intentions to expand participation from several hundred beta testers to thousands of authenticated security specialists over the next several weeks. Identity verification will remain mandatory for all access requests.
OpenAI initiated the integration of cybersecurity-specific protective measures into its systems throughout 2025. The company introduced Codex Security in March, a tool designed to detect and remediate software security flaws.
Regulatory Alarm Over Anthropic’s Model
Anthropic’s Mythos platform has sparked concern among senior American government officials. Bloomberg reported that Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell cautioned financial industry executives about potential Mythos-related threats last week.
Anthropic made Mythos available to approximately 40 major corporations to strengthen their cybersecurity infrastructure. The restricted distribution reflects the company’s concerns about unauthorized access to the technology.
On April 14, the director of Ireland’s National Cyber Security Centre testified before a parliamentary committee that threat actors are expected to deploy capabilities comparable to Mythos before year’s end.
Jack Clark, co-founder of Anthropic, predicted that competing organizations will introduce similar platforms within months. He further suggested that open-weight Chinese models with equivalent functionality could emerge within 12 to 18 months.
OpenAI’s Corporate Market Strategy
OpenAI has intensified its focus on business clients, a segment where Anthropic has rapidly established market presence. Ramp, a corporate payments platform, reported that approximately 33% of American companies purchased Anthropic solutions in March.
OpenAI has streamlined operations by discontinuing less lucrative initiatives. The organization suspended development of an adult-oriented ChatGPT version and halted the Stargate UK venture.
Microsoft, serving as OpenAI’s primary financial supporter, has committed to leasing data center infrastructure at a location initially designated for the Stargate Norway initiative. A separate agreement with British AI infrastructure provider Nscale collapsed.
OpenAI revealed plans to establish its first permanent London headquarters in 2027, accommodating over 500 employees. The facility will function as the company’s largest research center outside American borders.
The organization has also unveiled a new “superapp” concept influenced by Anthropic’s approach, and introduced specialized healthcare AI capabilities shortly following Anthropic’s release of its health-oriented Claude functionality.
