The UK has released new guidelines for secure AI system development in an effort to enhance cyber security in the field of AI. The National Cyber Security Centre (NCSC) collaborated with industry experts and 21 other international agencies and ministries, including the US Cybersecurity and Infrastructure Security Agency (CISA), to create the Guidelines for Secure AI System Development. These guidelines have been endorsed by 18 countries, including all G7 nations, making them the first globally agreed-upon guidelines of their kind.
The objective of these guidelines is to assist developers in making informed decisions regarding cybersecurity throughout the entire development process of AI systems. Whether the systems are built from scratch or based on existing tools and services, the guidelines aim to support developers in prioritizing cybersecurity. The official launch of the guidelines will take place at an event hosted by the NCSC, where industry leaders, government officials, and international partners will gather for a panel discussion on the common challenge of securing AI. Participants in the panel include Microsoft, the Alan Turing Institute, and cybersecurity agencies from the UK, US, Canada, and Germany.
Lindy Cameron, CEO of NCSC, described the guidelines as a significant step towards establishing a global understanding of cyber risks and mitigation strategies in AI development. She emphasized the importance of international collaboration between governments and industry to keep up with the rapid pace of AI advancement. Cameron had previously highlighted the risks of retrofitting security into AI systems in the future and stressed the need to integrate security measures during the development phase.
The new guidelines build upon the UK Government’s commitment to addressing AI risks through international cooperation. Prime Minister Rishi Sunak recently hosted the world’s first “AI Summit” in Bletchley Park, Buckinghamshire, where countries such as the UK, US, and China agreed on the necessity for international action to ensure safe and responsible AI development. This agreement, known as “The Bletchley Declaration,” has intensified discussions surrounding AI security measures.
Toby Lewis, Darktrace’s global head of threat analysis, emphasized that security is a prerequisite for trustworthy AI. He welcomed the guidelines’ emphasis on securing data and models from attackers and the importance of using the right AI for specific tasks. Lewis also highlighted the need for AI providers to build trust by transparently explaining how their AI systems arrive at their conclusions.
The release of these guidelines comes as the concerns over the rapid development of advanced AI systems continue to grow. Last year, hundreds of experts expressed their concerns in a letter issued by the Future of Life Institute, stating that advanced AI systems should only be developed once their positive effects and manageable risks are assured.
In line with its commitment to AI advancement, the UK government recently announced a £118 million boost for AI skills funding. This investment aims to ensure that the country possesses top global expertise and fosters the next generation of AI researchers. Additionally, the government has named 12 further Centres for Doctoral Training (CDT) in AI that will benefit from £117 million in government backing through UK Research and Innovation (UKRI). Furthermore, a new visa scheme will facilitate the entry of talented early-career AI researchers into the UK, making it easier for innovative businesses to attract top talent.
With the release of these guidelines and continued investment in AI, the UK is positioning itself as an “AI match-fit” nation, ready to harness the transformative benefits of AI technology.