The German government fears that a new artificial intelligence (AI) tool originating from the United States could drastically heighten the risks associated with catastrophic cyberattacks. The system, called “Mythos” is a product from the US firm Anthropic, and it is capable of detecting-and potentially exploiting-security loopholes on an unprecedented scale.
According to the newspaper Handelsblatt, citing government sources, the National Security Council has convened secretly to review the model and assess the implications of its potential spread. This council is tasked with coordinating the federal government’s security policy decisions, and therefore, the government has formally stated it will not comment.
The Federal Criminal Police Office (BKA) also recognizes significant inherent risks in these new applications. Carsten Meywirth, the BKA Director and head of the Cybercrime Department, told the Handelsblatt that while it is an initial positive development that vulnerabilities can be discovered and closed more quickly, he warned that cybercriminal methods and attack vectors adapt extremely fast to the state of the art. He emphasized that this dynamic-referring to specific ways hackers attack systems, such as through software flaws or manipulated emails-cannot be unilaterally halted.
In response, the federal government plans to hold talks with the US company. An Interior Ministry spokesperson confirmed to the Handelsblatt that, “The federal government is currently in dialogue with the manufacturer Anthropic”. When questioned about whether “Mythos” could contribute to the development of dangerous cyber weaponry, the spokesperson remained reserved, adding that the ministry “asks for understanding that it cannot make statements about possible implications at this time”.
