This comes after Anthropic’s Mythos sparked headlines across the country earlier this year about the risks it could pose to cybersecurity, going so far as to say it could not widely release the tech due to the threats it could create. The company reported that its Mythos-level technology could automatically identify security vulnerabilities in software.
In an email obtained by Government Technology, an offer for the credits was sent along with a survey that doubles as an application to participate. The survey identifies the program’s purpose as “to provide access to Claude cyber defense models and help train you to start triaging and patching systems with LLMs now.” In addition, it asks for one respondent per state, locality, tribe or territory, suggesting the program is available to other levels of government as well, committing up to $15M across the program, an Anthropic spokesperson confirmed.
In the email, Anthropic also encouraged state tech leaders to read a related blog, which notes that it is also making Claude Mythos 5 available early for a select group of “cyberdefenders and infrastructure providers.”
In simple terms, Anthropic appears to be giving some public-sector agencies at the state and local levels an opportunity to use its new AI products for free in order to prepare to tackle any related cybersecurity risks in the future.
“A credit amount of $100,000 per SLTT [state, local, tribal and territorial government] entity (expiring 6 months from provisioning date), to be used for scanning, triaging and patching systems with Claude Security, Claude Code + latest Opus 4.8 model with our cyber runbooks, capped at 100 total states, tribes, largest counties/cities,” the company wrote in the email.
Anthropic has also announced plans to ultimately release Mythos 5 through Project Glasswing, an initiative that the company created to put its tech to work for cyber defense.
Editor's Note: This story has been updated with comment and clarification from Anthropic.
