Government security experts are warning that cyber criminals could gain the ability to steal from bank accounts using artificial intelligence within the next three months. Officials working with Britain's AI Security Institute (AISI) have expressed fears of a "nightmare scenario" emerging due to rapidly advancing AI technology, particularly models developed in China that reportedly lack stringent safety guardrails.
These AI models, unlike Western counterparts, allow developers to more easily disable safety features that restrict potentially harmful uses. This could enable a widespread availability of tools capable of breaking advanced passwords, stealing personal financial information, and committing fraud. Insiders suggest that hackers from various regions could soon access this technology.
The Financial Stability Institute has also alerted financial institutions, stating that some current AI technology is so advanced that it may leave banks with only hours to stop a breach. Ministers are reportedly concerned about the pace of AI development and anticipate a significant increase in sophisticated cyberattacks.
Concerns about AI safety have been heightened by recent actions from major AI firms. Anthropic, a prominent AI company, reportedly barred Britain's AI Security Institute from testing its latest model, which some interpret as AI being treated as a sensitive national security issue akin to "dangerous weapons." Anthropic stated its AI model, Claude, has the capability to assist in scientific breakthroughs but also carries the risk of being used for malicious purposes, citing five instances this year where the model was allegedly used in attempts to create bioweapons. The company's reports indicate that the same biological research capabilities could be exploited to engineer dangerous pathogens.
These developments follow a global debate ignited by a former Anthropic researcher's warning about an "out of control" AI arms race and the potential for AI to pose an existential threat to humanity. AI safety advocates are calling for governments to take the risks associated with AI more seriously.
In response to these concerns, a government spokesperson stated that while AI is enhancing certain cyber capabilities, businesses are not without recourse. The spokesperson emphasized the importance of businesses urgently improving their cyber defenses and highlighted practical steps such as adopting Cyber Essentials, following National Cyber Security Centre guidance, and prioritizing cybersecurity at the board level.