OpenAI restricts access to Cyber after limiting Mythos
▼ Summary
– Sam Altman confirmed OpenAI will restrict access to its cybersecurity tool Cyber, similar to Anthropic’s approach with Mythos, after criticizing that company for gatekeeping.
– OpenAI is rolling out GPT-5.5 Cyber to critical cyber defenders in the coming days, with an application process for access.
– Cyber can perform penetration testing, vulnerability identification, exploitation, and malware reverse engineering to find security holes.
– Altman previously called Anthropic’s restriction of Mythos fear-based marketing, but an unauthorized group reportedly accessed Mythos anyway.
– OpenAI plans to make Cyber more widely available by consulting with the U.S. government and verifying legitimate cybersecurity credentials.
OpenAI has confirmed it will follow a similar path to Anthropic by restricting access to its new cybersecurity tool, Cyber , a move that comes just weeks after CEO Sam Altman publicly criticized Anthropic for doing the same with its own tool, Mythos.
In a post on X Thursday, Altman announced that OpenAI would begin rolling out GPT-5.5 Cyber “to critical cyber defenders” in the coming days. To get access, users must submit an application through OpenAI’s website, providing details about their credentials and intended use cases. The company is essentially gatekeeping the tool behind a verification process.
According to the application page, Cyber is designed to handle a range of high-stakes cybersecurity tasks, including penetration testing, vulnerability identification and exploitation, and malware reverse engineering. It’s meant to serve as a comprehensive toolkit for organizations looking to find and fix security weaknesses. But the same capabilities that make it valuable for defenders also raise the risk of misuse by malicious actors.
Altman’s decision to restrict access is a notable reversal. When Anthropic limited access to its own cybersecurity tool, Mythos, Altman dismissed the approach as fear-based marketing. Critics agreed, arguing that Anthropic’s warnings were overblown. In an ironic twist, reports emerged that an unauthorized group had managed to gain access to Mythos anyway, undermining the very restrictions Anthropic had put in place.
OpenAI says it is working to eventually make Cyber more widely available. The company is consulting with the U. S. government and actively identifying additional users with legitimate cybersecurity credentials to expand access over time. For now, however, only a select group of defenders will get their hands on the tool.
(Source: TechCrunch)
