Hackers Could Exploit AI Tool ChatGPT to Target UK’s NHS and MoD, Oxford Professor Warns
An Oxford professor has issued a stark warning about the potential risks of AI tool ChatGPT, stating that hackers could exploit its capabilities to launch cyber attacks on the National Health Service (NHS) and the Ministry of Defence (MoD). Michael Wooldridge, a computer science professor at Oxford University, highlighted that criminals could use ChatGPT to rapidly produce computer programs capable of overwhelming government websites, thus triggering crashes and impairing essential services.
The specific concern lies in the ability of ChatGPT and similar large language models to provide uncensored access to computer programming skills. These models possess no ethical principles whatsoever, Wooldridge cautioned during his speech at the Cheltenham Literature Festival. He outlined the potential danger of individuals misusing AI tools like ChatGPT to launch denial of service (DoS) attacks, which involve flooding a website with an overwhelming amount of traffic to cause a crash.
Wooldridge went on to explain that traditional DoS attacks would typically require programmers to write the malicious code, and most ethical programmers would refuse to participate in such activities. However, with the advent of large language models like ChatGPT, anyone with access to such a tool would be empowered to generate 20-30 lines of code in a matter of minutes. While the creators of ChatGPT, OpenAI, would likely attempt to prevent the tool’s misuse for cyber attacks, Wooldridge raised concerns about the potential dangers posed by other AI tools in the future.
The professor expressed worry about a scenario in which numerous individuals possess access to large language models downloaded onto their personal devices, enabling unchecked access to coding capabilities. He stressed that the absence of barriers or guardrails in this context could lead to significant harm, especially through cyber attacks. Wooldridge emphasized that cyber attacks have become highly prominent concerns, and governments, including the UK, are fundamentally worried about the potential risks associated with this form of targeted aggression.
The warning comes just ahead of an international AI safety conference hosted by UK Prime Minister Rishi Sunak. Set to take place at Bletchley Park, the historical site of British code-breaking operations during World War II, the event will convene representatives from various governments, including China, as well as tech companies and academics. The conference aims to explore the risks posed by AI technology and how to effectively manage and mitigate these risks.
In light of Professor Wooldridge’s cautionary remarks and the pressing need to address the potential weaponization of AI, the global summit could kick-start vital discussions and collaborations to ensure the responsible development and utilization of AI tools, while safeguarding against cyber threats. As AI continues to advance, it is crucial to strike a balance between innovation and security to protect critical systems and institutions.
By bringing together diverse stakeholders, such as governments, tech companies, and experts, this conference presents an opportunity to collectively navigate the complexities of AI safety and take proactive measures to prevent malicious actors from exploiting such powerful tools. As technology progresses, it is vital to remain vigilant and continuously reassess our approach to AI development in the pursuit of a safer and more secure digital landscape for all.
