‘This version of GPT-5.4 lowers the rejection threshold for ‘legitimate work’ in cybersecurity’, OpenAI said.
OpenAI said it will only allow select verified users access to its latest AI model for cybersecurity operations, a week after the limited launch of Anthropic’s Mythos.
Purpose-built for security operations, the new GPT-5.4-Cyber will be available to users willing to work with OpenAI to certify themselves as cybersecurity defenders, the company said.
This version of GPT-5.4 lowers the threshold for rejecting “legitimate” cyber security work. As a “highly permissive” model, OpenAI said it is starting to ship GPT-5.4-Cyber to “vetted” security vendors, organizations, and researchers.
The maker of ChatGPT only started to integrate Internet-specific protection into its implementation model from 2025, and introduced Codex Security to identify and fix vulnerabilities in March. In February, it launched Trusted Cyber Access as a way to verify the identity of cybersecurity professionals.
The new Mythos model of Anthropic shows the important capabilities of finding and producing safety equipment. Worried about bad actors, Anthropic made the choice to offer Mythos to a group of 40 large enterprises to strengthen their cyber defenses.
The reported ability of Mythos has already raised concerns among world leaders. Yesterday (14 April), the director of the National Cyber Security Center told the Oireachtas Joint Committee on AI that more models like Mythos should be expected in the hands of bad actors before the end of the year.
Anthropic founder and policy leader Jack Clark had similar beliefs. “There will be other programs like this in a few months from other companies, and then after a year to a year and a half, there will be open-weight models from China with these capabilities,” he told the audience at the Semafor World Economy event in Washington DC earlier this week.
OpenAI, which plans to offer its first public offering later this year, has been trying to focus on the enterprise market — a sector that Anthropic is quickly taking on. According to data from payments group Ramp, nearly one in three US businesses paid for Anthropic tools in March.
The company has been putting off projects that don’t make much money, including “permanently” halting plans for the sensational ChatGPT and stopping Stargate UK.
On the other hand, the main supporter of OpenAI Microsoft, has agreed to rent the capacity of the data center on the site intended for the Stargate Norway project, since one of the OpenAI agreements with the UK AI infrastructure Nscale failed to start.
Competition between the two companies has intensified, with the announcement of a new Anthropic-inspired ‘superapp’ by OpenAI, or a dedicated set of AI health tools by Claude launched just days after OpenAI released ChatGPT Health.
Apart from suspending plans for Stargate UK, OpenAI said it is opening its first permanent office in London in 2027 with more than 500 people. The company plans to make London its largest research center outside the US, it said.
Don’t miss out on the information you need to succeed. Sign up for Daily BriefSilicon Republic’s digest of must-know sci-tech news.
