Economic authorities in the United States and the Financial Market Commission (CMF) in Chile are monitoring Anthropic's new AI model, Claude Mythos, over potential cybersecurity risks. The model has triggered alarms due to its advanced engineering capabilities in software and security.
Rolando de la Cruz, an academic at the Universidad Adolfo Ibiente (UAI) Faculty of Engineering and Sciences, warned that the AI can facilitate rapid and sophisticated cyberattacks. He told Expreso Bío Bío that the technology could lead to an escalation of damage within critical infrastructure.
Offensive Capabilities
De la Cruz noted that while the technology possesses dual-use potential for both attack and defense, its offensive capacity currently outpaces existing security measures. "Claude Mythos has an advanced offensive capacity in cybersecurity, faster than defenses," De la Cruz said.
The AI model can identify vulnerabilities within a system and generate rapid actions to exploit them. This capability significantly lowers the barrier to entry for malicious actors.
"In the past, a person wanting to dedicate themselves to hacking needed much more advanced knowledge of programming systems. Now, they don't," De la Cruz added.
Due to these risks, Anthropic has restricted access to the general public. Currently, the model can only be accessed via invitation. The company's research and security divisions have implemented these limitations to manage the potential for large-scale system breaches.
