Over the past weeks, various technology experts and financial leaders have raised concerns over the potential misuse of Anthropic’s new powerful AI model: Cloud Mythos Preview.
Now that their fears have been justified bloomberg The Mythos AI model, equipped with unprecedented cyber capabilities, has been accessed by a small group of unauthorized users, reports said on Wednesday.
According to a source close to the matter, when Anthropic decided to release the model to some tech companies under Project Glasswing, a handful of users connected to a private online forum gained access to Mythos.
What is more worrying is that since then unauthorized users have been using Mythos for various purposes.
An Anthropic spokesperson said, “We are investigating a report claiming unauthorized access to Claude Mythos Preview through one of our third-party vendor environments.”
Since its release, Anthropic has introduced Mythos AI as a powerful model that can identify and exploit critical vulnerabilities in every “major operating system and web browser” when directed by the user. Given its capabilities, the US-based artificial intelligence company has decided to scale back its public-wide release.
If misused, Mythos could pose a serious threat to economies, public safety, and national security, the Anthropic CEO warned.
In recent days, heads of major financial institutions and the banking system have expressed concerns over the risks posed by AI models. They are also urging people to join the list of early testers to protect their own systems.
Additionally, recent reports of unauthorized access highlight security flaws and highlight the challenges Anthropic faces in completely preventing access to its most powerful models beyond approved partners.
