OpenAI Extends Cyber Program to Government Agencies
▼ Summary
– OpenAI published a roadmap to expand its Trusted Access for Cyber (TAC) program to governments at all levels, focusing on US and allied agencies.
– The TAC program automates identity verification for cybersecurity defenders and now includes tiers exclusive to authenticated defenders.
– OpenAI released GPT5.4-Cyber, a large language model aimed at helping cybersecurity professionals.
– The expansion covers federal, state, and local government missions, including national security, public health, and critical infrastructure.
– This effort follows Anthropic’s conflict with the US government, which labeled the company a supply chain risk over military access to its AI models.
OpenAI has unveiled a strategic roadmap aimed at “democratizing AI-powered cyber defense,” pledging to extend its Trusted Access for Cyber (TAC) program to government agencies “at every level.” The announcement, detailed in a document titled Cybersecurity in the Intelligence Age, was released by OpenAI’s Global Affairs team on April 30 , just days after the company launched its latest large language model for cybersecurity defenders, GPT5.4-Cyber.
A central commitment in the plan is broadening TAC program access for cyber defenders across federal, state, and local government entities. This includes support for national security missions, threat response, public health systems, emergency management, benefits delivery, and local critical infrastructure. While OpenAI did not name specific governments set to join the program, the document’s introduction explicitly references the “US and its allies,” strongly suggesting that US government agencies could be among the first participants.
Initially introduced in February, the TAC program was designed to automate identity verification, reducing friction in safeguards for cybersecurity tasks while partnering with a select group of organizations. An updated version of the program was announced alongside GPT5.4-Cyber, featuring new tiers , including one reserved exclusively for “users willing to work with OpenAI to authenticate themselves as cybersecurity defenders.”
This push for closer government collaboration comes as Anthropic, one of OpenAI’s top competitors, faces a tense standoff with the Trump administration. In March, the administration designated Anthropic a “supply chain risk” after the company refused to grant the US military unrestricted access to its Claude AI models. Anthropic has since launched Claude Mythos Preview, an LLM reportedly capable of autonomously finding and fixing cybersecurity vulnerabilities at scale, as well as Project Glasswing, an initiative aimed at discovering and patching software vulnerabilities using LLMs.
(Source: Infosecurity Magazine)
