Anthropic Confirms Claude Mythos AI Model After Major Data Leak
Anthropic acknowledges testing a powerful new AI model called Claude Mythos after a security lapse exposes thousands of unpublished documents. The company calls the model its most capable to date despite unprecedented cybersecurity risks.
Anthropic confirms it is developing and testing a powerful new AI model called Claude Mythos after a data leak accidentally reveals its existence. A company spokesperson describes the new model as a major step change in AI performance and the most capable system the lab has built to date. Anthropic currently provides the model to a select group of early access customers for trials.
The details emerge because Anthropic inadvertently leaves nearly 3,000 unpublished blog assets in an unsecured, publicly searchable data store. Cybersecurity researchers discover this cache, which includes a draft blog post detailing Claude Mythos and a new, more intelligent tier of models named Capybara. The draft warns that this new technology poses unprecedented cybersecurity risks.
After Fortune informs Anthropic of the leak, the company secures the data store and blocks public access. Anthropic blames the exposure on a human error in the configuration of its content management system and characterizes the compromised files as early drafts considered for publication. The leak also reveals plans for an invite-only CEO summit in Europe aimed at selling AI models to large corporate clients.