After Sam Altman trash-talked Anthropic for gatekeeping its cybersecurity instrument Mythos by solely releasing it to pick out customers, he confirmed that OpenAI could be doing the identical with its competing instrument, Cyber.
Altman mentioned in a put up on X on Thursday that OpenAI will start rolling out GPT-5.5 Cyber “to vital cyber defenders” within the subsequent few days. OpenAI has an software on its web site the place individuals submit details about their credentials and deliberate use with a view to achieve entry.
This model of Cyber can carry out such duties as penetration testing, vulnerability identification (and exploitation), and malware reverse engineering, the appliance implies. It’s meant to be a toolkit to assist an organization discover safety holes and take a look at defenses. The concern is that the package might be misused by the dangerous guys.
When Anthropic equally restricted entry to Mythos, Altman referred to as the tactic fear-based advertising. Some critics additionally thought so, saying Anthropic’s rhetoric was overblown. Sarcastically, an unauthorized group reportedly managed to achieve entry to Mythos anyway.
OpenAI says it’s working to make Cyber extra extensively obtainable by consulting with the U.S. authorities and figuring out extra customers with legit cybersecurity credentials.
A spokesperson tells Trendster that the corporate’s system for verifying these with legit cybersecurity credentials, which it calls Trusted Entry for Cyber (TAC) has scaled “to hundreds of verified defenders and a whole lot of groups chargeable for defending vital software program.” These people can use the newest mannequin, GPT 5.5 for “cybersecurity duties” with much less “friction” from safeguards.
The TAC permissions program is tiered, the spokesperson mentioned: “Vital defenders with reputable defensive use instances can apply to entry devoted extra cyber-permissive fashions like GPT 5.4-Cyber, and the forthcoming GPT 5.5-Cyber, by this system.”
Be aware: This story was up to date to incorporate an announcement from OpenAI.
