It has the ability to act autonomously as an agent, meaning that it can simultaneously deal with software environments, browse the internet, and scan complex codebases.
Mythos could act as an incredible defensive tool for identifying and patching holes to keep banks, governments, and corporations safe. In the hands of bad actors, however, Mythos could act as a highly sophisticated hacker.
Anthropic has severely restricted its initial rollout due to the catastrophic potential of releasing Mythos into the wild.
The AI firm launched "Project Glass Wing," a coalition granting early preview access to select tech companies, government organizations, and traditional Wall Street banks.
However, major cryptocurrency companies were notably excluded from this early access program.
Firms are urgently lobbying Anthropic for access so they can preemptively scan their own systems .
Coinbase Chief Security Officer Philip Martin recently confirmed that the exchange is currently in "close communication" with Anthropic regarding Mythos.
When Mythos goes live, hackers could turn their sights on the blockchain.