Is this just “fear-based marketing”?
The new results from GPT-5.5 show that, when it comes to cybersecurity risk, the Mythos preview was likely not “a breakthrough specific to one model,” but rather “a byproduct of more general improvements in long-horizon autonomy, logic, and coding,” AISI writes.
In a recent interview with the Core Memory podcast, OpenAI CEO Sam Altman criticized “fear-based marketing” in promoting limited releases for some AI models. Although he said he was “sure that Mythos is a great model for cybersecurity,” he added that “It’s obviously incredible marketing to say, ‘We’ve built a bomb. We’re going to drop it on your head. We’ll sell you a bomb shelter for $100 million.'”
Altman added, “There will be a lot of rhetoric about models that are too dangerous to release.” “There will also be very dangerous models that will have to be released in different ways.”
In February, OpenAI launched its Trusted Access to Cyber Pilot program, allowing security researchers and enterprises to verify their identities and register their interest in studying OpenAI’s Frontier models for “legitimate defensive work.” Last month, OpenAI said it was using that trusted access list to moderate the limited launch of GPT-5.4-Cyber, a model version that it says has been “intentionally fine-tuned for additional cyber capabilities and with fewer capacity restrictions.”
On Thursday, OpenAI CEO Sam Altman said on social media that the initial release of GPT-5.5-Cyber will similarly be limited to “critical cyber defenders over the next few days.”
<a href