Anthropic's Mythos AI model, withheld from public release due to cybersecurity risks, threatens share prices of Okta, JPMorgan Chase, and Coinbase.

Anthropic announced that Mythos, its latest AI model, is too capable to release publicly because it can find and exploit cybersecurity vulnerabilities. The company stated that 'The fallout -- for economies, public safety, and national security -- could be severe.'

Okta provides identity and access management (IAM) services that help customers control who can access files and data. The company has more than 20,000 customers and its share price has fallen about 24% over the past year.

JPMorgan Chase CEO Jamie Dimon stated that new AI models like Mythos make cybersecurity 'worse' and that the model revealed 'a lot more vulnerabilities need to be fixed.' The U.S. Treasury Department and Federal Reserve recently met with CEOs of nearly all major U.S. banks to discuss Mythos.

Coinbase has 9.2 million monthly transacting users and about $376 billion in assets on its platform. More than 40 companies have already received access to the preview version of Mythos to help them find and fix vulnerabilities.