Anthropic is releasing its new model to select companies to get a head start on cyber defense
Fortune reported that the Claude developer would be doing this in late March.
Anthropic said that it’s releasing “Mythos Preview,” a version of its new AI model, to a select group of mostly tech companies (and JPMorgan) so they can get to work on protecting against cyberattacks.
Shares of CrowdStrike and Palo Alto Networks, two cybersecurity firms on this list, caught a bid on the news.
Anthropic dubbed this initiative “Project Glasswing,” with the goal to “secure the world’s most critical software.”
Those cybersecurity stocks had faced heavy selling pressure in late March, when a leaked Anthropic document reviewed by Fortune warned that its new model would be so powerful that malicious actors could launch potentially indefensible cyberattacks.
That same report also said that Anthropic would be releasing Mythos early to cybersecurity firms to help them bolster their defenses... which is exactly what Project Glasswing is doing.
It’s been a busy 24 hours for Anthropic, with the Claude developer announcing an expansion of its partnership with Google and Broadcom for AI compute and saying that its run-rate revenue has surpassed $30 billion, up from about $9 billion at the end of 2025.
Some of that jump could be juiced by Meta’s engineers, whose AI usage may be serving as another example of Goodhart’s Law (per this report from The Information). That being said, there’s clearly some breadth, and not just depth: Anthropic said the number of business customers spending over $1 million on an annualized basis has doubled since February to above 1,000.
As an aside, sincere congratulations to JPMorgan and CEO Jamie Dimon for effectively being recognized as a tech company, something every bank has wanted to be known as for at least a decade. Per Anthropic’s press release:
