Key Highlights
- GPT-5.4-Cyber has been made available to a vetted group of cybersecurity professionals through OpenAI’s verification program
- Anthropic unveiled Mythos, its security-focused AI system, seven days prior to OpenAI’s announcement
- U.S. Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell expressed concerns regarding Anthropic’s Mythos
- OpenAI’s roadmap includes scaling access from several hundred to multiple thousand verified security practitioners
- Ramp’s business spending data revealed nearly 33% of American companies purchased Anthropic services during March
OpenAI introduced a specialized AI system tailored for cybersecurity operations. GPT-5.4-Cyber represents the company’s latest offering, currently deployed among a carefully selected group of 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 Tuesday, April 15 debut arrived seven days after Anthropic’s controlled rollout of Mythos, its competing security-oriented system.
According to OpenAI, GPT-5.4-Cyber operates as an enhanced variant of GPT-5.4 with relaxed restrictions. The model features adjusted safeguards that permit authorized defensive cybersecurity operations.
Current availability remains restricted to members of OpenAI’s Trusted Access for Cyber initiative. This program, which debuted in February, mandates rigorous identity authentication for all participating users.
The company announced expansion plans to broaden program participation from its current hundreds of participants to several thousand authenticated defenders. Identity verification requirements will remain mandatory for all future access grants.
Throughout 2025, OpenAI has integrated specialized cybersecurity protections into its AI systems. March saw the debut of Codex Security, a tool designed for detecting and resolving software security flaws.
Regulatory Scrutiny Surrounding Mythos
Anthropic’s Mythos platform attracted scrutiny from senior government officials shortly after release. Bloomberg reported that Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell cautioned financial industry executives about potential risks associated with the model.
Anthropic made a deliberate decision to distribute Mythos among approximately 40 major corporations for defensive security purposes. This restricted distribution strategy stemmed from concerns about potential misuse by malicious actors.
During an April 14 parliamentary hearing, Ireland’s National Cyber Security Centre director projected that threat actors would develop comparable capabilities before year’s end.
Jack Clark, Anthropic’s co-founder, predicted competing companies would deploy similar technologies within the coming months. He projected that Chinese developers might release open-weight equivalents within 12 to 18 months.
Enterprise Market Competition Intensifies
OpenAI has shifted strategic focus toward corporate clients, a segment where Anthropic has established significant presence. March statistics from payments platform Ramp indicated that approximately 33% of American enterprises subscribed to Anthropic’s services.
Recent moves show OpenAI trimming initiatives with lower profit margins. Development of an adult-oriented ChatGPT variant has been suspended, while the Stargate UK infrastructure project faces indefinite delays.
Microsoft, serving as OpenAI’s primary financial supporter, secured a lease agreement for data center infrastructure at a location initially designated for Stargate Norway. A separate partnership with British AI infrastructure provider Nscale was terminated.
OpenAI revealed plans for establishing its inaugural permanent London headquarters in 2027, designed to accommodate over 500 employees. The facility will serve as the company’s primary research center outside American borders.
Recent product announcements from OpenAI include a comprehensive “superapp” platform drawing inspiration from Anthropic’s approach, along with specialized healthcare AI capabilities released shortly after Anthropic deployed its Claude health features.
