Anthropic’s Mythos artificial intelligence model gained access to most of the US National Security Agency’s systems during testing, raising serious concerns within US security circles.
- Mythos has been recognised as one of the most powerful AI systems in cybersecurity.
- The system has demonstrated the ability to identify vulnerabilities in computer networks.
- The US has restricted access to the Mythos 5 and Fable 5 models for foreign nationals.
- Anthropic has suspended access to its models following government intervention.
Why Mythos caused concern
Model testing results Mythos impressed representatives of the US intelligence services. The system demonstrated the ability to automatically search for vulnerabilities, analyse secure networks and create complex scenarios for infiltrating information systems. These capabilities raised serious concerns about the potential threats Mythos might pose.
«We cannot ignore the risks associated with this technology,» said a US government representative.
Restrictions on access to models
In June, the US administration decided to impose export restrictions on the Mythos 5 and Fable 5 models. The government expressed concerns that these technologies could be used by foreign intelligence services to carry out cyberattacks. Upon receiving the relevant directive, the company Anthropic has suspended access to the models for users worldwide, whilst working with the US authorities to address the risks.
Controversy surrounding the authorities’ decision
The restrictions have drawn criticism from some cybersecurity experts. More than a hundred specialists have signed an open letter calling for the ban to be reviewed, emphasising that such models can be useful for protecting critical infrastructure and identifying vulnerabilities in software. However, the US authorities insist that the restrictions are necessary due to potential threats to national security.
