Mythos, a new AI model from Anthropic, is triggering growing concern across the banking sector amid warnings it could accelerate and scale complex cyberattacks. The concern stems from the model’s advanced coding and autonomous capabilities, giving it an unprecedented ability to detect and exploit security vulnerabilities at high speed.
Details
Anthropic launched the model on April 7, describing it as its most advanced system for programming and autonomous operations.
- Cybersecurity experts say the model can analyze complex technical architectures, including legacy systems that still form the backbone of banking infrastructure.
- Banks rely on a mix of modern and decades-old systems, creating an environment rich in undiscovered vulnerabilities.
- High interconnectivity between financial institutions, and shared reliance on vendors and technologies, increases the risk of large-scale breach propagation.
During testing phases, the model identified thousands of critical vulnerabilities, including long-standing flaws in widely used software and errors in sensitive virtual systems. Experts warn these capabilities could lower the cost of attacks and reduce the need for advanced skills, expanding the threat landscape.
In response, governments in the United States, Canada, and the United Kingdom have begun discussions with banking leaders to assess risks. Anthropic has also launched a closed evaluation program involving major tech companies and financial institutions to develop proactive defenses.
What’s Next?
Attention is shifted to the results of ongoing tests and whether banks can modernize systems and deploy defenses capable of keeping pace with a new generation of AI-driven threats.
