Anthropic's neural network has prompted British banks to strengthen their cybersecurity — FT
UK financial regulators have launched emergency consultations with the government’s cybersecurity agency and leaders in the banking sector.
The FT reports this.
The cause for concern is Anthropic’s new artificial intelligence model, Claude Mythos, which could pose a threat to the stability of critical IT infrastructure.
Representatives from the Bank of England, the Financial Conduct Authority, and the Treasury are at the center of the discussions.
Together with the National Cyber Security Centre, they are examining the vulnerabilities in systems that this neural network is capable of detecting and exploiting.
It is expected that over the next two weeks, key banks, insurance companies, and stock exchanges will receive detailed instructions on how to counter these new cyber risks.
The new model demonstrated such high effectiveness that the developers decided against its public release.
Instead of public access, “Project Glasswing” was initiated—a closed-door testing program in a controlled environment.
Tech giants such as Apple, Google, Nvidia, and the largest U.S. bank, JPMorgan Chase, were brought on board to participate in the project.
The main cause for concern was that in just a few weeks of operation, Mythos discovered thousands of “zero-day” vulnerabilities.
These flaws in operating systems and browsers were previously unknown to developers, making them ideal targets for hacker attacks.
The U.S. government held an emergency meeting with the heads of the largest banks over potential cyber threats related to Anthropic’s new AI model.
The ShinyHunters hacker group attacked the developers of Grand Theft Auto 6
