"Anthropic has trained a new AI model, called 'Mythos', which the company claims represents a 'step change' in capabilities. However, the company is particularly worried about the cybersecurity risks the model poses."
"Anthropic inadvertently leaked sensitive documents, including a draft blog post about Mythos, in an unsecured database, raising alarms about the security of their internal information."
"The new cybersecurity capabilities of AI models have become so concerning that both Anthropic and OpenAI have informed the government about the dangers of their developing models."
"Anthropic has suffered another major security lapse, leaking the code of the agentic harness that surrounds Claude Code, which could have serious implications."
Anthropic has trained a new AI model named Mythos, which is said to represent a significant advancement in capabilities. However, the company has experienced multiple data leaks, including sensitive documents and code related to the model. These leaks raise concerns about cybersecurity risks associated with AI. Both Anthropic and OpenAI have alerted the government about the potential dangers of their AI models, providing early access to security experts to address these issues.
Read at Fortune
Unable to calculate read time
Collection
[
|
...
]