A group of unauthorized users used contractor credentials and guessed URL patterns to access Anthropic’s restricted ‘Claude Mythos’ model, involving a hacked AI, leading the company to investigate the report stating this incident.
Keeping advanced technology secure, especially when multiple vendors are involved, is a challenging matter. While companies try to use these powerful systems to strengthen digital defenses and permit government to have access, the reality of unauthorized AI access shows that maintaining perfect control is a difficult goal in today’s fast-moving tech world.
Anthropic’s AI Being Hacked Scandal
The situation unfolded on the same day Anthropic introduced its ‘Project Glasswing’ initiative, a program meant to provide a select group of major tech companies with access to the model.
However, a small group of people managed to reach the model through a third-party vendor environment. By using its credentials, they were able to bypass typical security steps, which is why experts say that when a company hacks AI model processes, the weakest link is often the supply chain.
Because of these events, news regarding Mythos AI hacked concerns has spread quickly through the industry.
“We’re investigating a report claiming unauthorized access to Claude Mythos Preview through one of our third-party vendor environments,” said an Anthropic spokesperson.
Meanwhile, the company clarified there is no evidence that the AI hack went beyond the specific vendor space or reached its core systems. This distinction is fundamental for investigators trying to understand how such a sensitive piece of software could be exposed.
Technology Changing the Game
The concern here is the nature of the tool, as Mythos can discover thousands of software vulnerabilities and write working code to exploit them. Security experts are worried about the potential for adversarial attacks on AI if such tools fall into the wrong hands.
In this case, reports of the Claude AI hack have put security leaders on high alert regarding their own infrastructure. Ram Varadarajan, CEO of Acalvio Technologies, noted that the event did not require a complex effort or advanced technical skills.
“It just required a contractor, a URL pattern and a Day-One guess, which means the ‘controlled release’ model failed at its weakest link, before the model’s capabilities were ever the issue” he said.
This simple method led to hacked AI, proving that even a well-intentioned AI hack can create significant industry alarms.
The worry falls on governments and business that have adopted this model, especially the UK business that are now wondering how they can prevent their Anthropic’s AI being hacked in the future.
As news of the Mythos AI hacked incident continues to circulate, professionals are re-evaluating their own security measures and internal policies. While the group involved claimed their intentions were just to play around with tech, the threat remains real for the broader cybersecurity community.
As head of software supply chain risk strategy at Black Duck Software, Tim Mackey, noted, for defenders, even the thought of an ‘hacks AI model scenario’ increases stress and highlights vulnerabilities.
“Anthropic’s marketing message for Mythos was effectively a challenge,” Mackey claimed.
Ultimately, because of the risk, the Mythos AI hacked report serves as a wake-up call for how we manage powerful software. Security leaders everywhere must now consider how to defend their operations against hacked AI attempts and ensure that their systems are truly protected.
It is clear that the industry is still learning how to balance progress with safety, and a hacked AI serves as a stark reminder of that ongoing journey. Governments and businesses must remain cautious, as guarding against a hacked AI is essential to keeping the digital future safe from unintended consequences.
Inside Telecom provides you with an extensive list of content covering all aspects of the tech industry. Keep an eye on our Cybersecurity sections to stay informed and up-to-date with our daily articles.