Key Highlights
- GPT-5.4-Cyber debuted for a carefully selected group of authenticated cybersecurity experts
- The announcement arrived seven days following Anthropic’s security-oriented AI system, Mythos
- Federal officials including Treasury Secretary Scott Bessent and Fed Chair Jerome Powell expressed caution about Mythos
- OpenAI will broaden availability from several hundred to several thousand authenticated security practitioners soon
- Ramp data shows approximately 33% of American companies purchased Anthropic’s products during March
OpenAI introduced a specialized AI system designed exclusively for cybersecurity applications. The system, designated GPT-5.4-Cyber, became accessible to a carefully selected group of authenticated professionals.
The company made the announcement on Tuesday, April 15. Anthropic had introduced its security-focused system, Mythos, with restricted availability exactly seven days earlier.
GPT-5.4-Cyber represents a modified variant of GPT-5.4 with expanded permissions. The system features reduced restrictions for cybersecurity operations that OpenAI classifies as authorized defensive activities.
Current availability remains restricted to members of OpenAI’s Trusted Access for Cyber initiative. This initiative began in February and mandates identity authentication for professionals seeking model access.
OpenAI outlined intentions to scale the initiative from several hundred participants to several thousand authenticated defenders over the next few weeks. The organization confirmed that identity verification will remain mandatory for access approval.
OpenAI introduced cyber-focused protective measures to its systems starting in 2025. During March, the company unveiled Codex Security to assist in detecting and resolving software security flaws.
Federal Attention on Anthropic’s Mythos
Anthropic’s Mythos system attracted scrutiny from senior federal officials. During the previous week, Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell advised Wall Street executives to recognize potential threats from Mythos, Bloomberg reported.
Anthropic selected approximately 40 major corporations to receive Mythos access for strengthening their cybersecurity infrastructure. The company maintained tight access controls based on worries about potential misuse.
The director of Ireland’s National Cyber Security Centre informed a parliamentary committee on April 14 that threat actors could possess comparable capabilities to Mythos before year-end.
Anthropic co-founder Jack Clark predicted rival companies would introduce comparable systems in the coming months. He projected that open-weight Chinese models featuring these capabilities might emerge within 12 to 18 months.
Enterprise Market Focus
OpenAI has intensified efforts targeting enterprise clients, a segment where Anthropic has experienced rapid expansion. Payment processor Ramp’s data indicates approximately 33% of American businesses subscribed to Anthropic products during March.
OpenAI has reduced investment in lower-margin initiatives. The organization suspended development of an adult-oriented ChatGPT variant and delayed its Stargate UK initiative.
Microsoft, serving as OpenAI’s primary investor, committed to leasing data center infrastructure at a location initially designated for the Stargate Norway initiative. A separate agreement with British AI infrastructure company Nscale ultimately collapsed.
OpenAI revealed plans for establishing its first permanent London headquarters in 2027, accommodating over 500 personnel. The company identified London as its largest research center beyond American borders.
OpenAI recently unveiled a new “superapp” following Anthropic’s example, while introducing specialized AI healthcare solutions shortly after Anthropic deployed its health-oriented Claude capabilities.
