New UK Government measures can deliver a global framework to strengthen the protection of AI models against hacking and fraudulent activities.

The new measures are believed to support the development of a global standard on how to enhance the protection of AI models from hacking, and were announced by the UK Government on the 15th of May.  During a discussion at CYBERUK, a dedicated cyber security conference, the technology minister, Saqib Bhatti, announced two new measures supporting developers to strengthen cyber security in AI models and software. The hope is that these practices will enable the UK economy to progress safely and support the Government in reaching long-term economic growth.

The codes include requirements for developers to make products resilient against possible hacking or sabotage and improve confidence in applying AI models, support business efficiency, drive growth and accelerate innovation. In the last year, government reports suggest that about half of companies and a third of charities experience cyber attacks, and phishing remains the most popular form of security breach. The codes introduced will assist developers in creating software more securely to prevent attacks similar to the one witnessed with the MoveIT software in 2023.

Saqib Bhatti explains that to utilise the significant potential of the digital economy, we must deliver safe conditions to enable it to grow and develop. These new measures will allow this to happen, making AI models more resilient in the design stage.

The latest report shows the UK is committed to making the economy more resilient to potential attacks and delivering success and opportunities to communities nationwide. It’s critical to see continued growth in the industry, solidifying the UK as a global leader in the cyber security market. 

The new measures come when recent findings indicate the cyber security industry has experienced a 13% growth and is now valued at nearly £12 billion, matching other markets such as the automotive industry. 

The new codes of practice will enhance cyber security in AI and software, while new government processes on cyber skills will support the progression of the cyber workforce and ensure the nation has the suitable people to protect the UK online.

Felicity Oswald, the CEO of NCSC, explains that to harness the most of technology progression, cyber security must be at the core of how we build digital systems. Creating standards for our national security will deliver collective resilience, and Oswald has praised businesses that follow these standards and help maintain the safety of the UK online. These steps are critical for new companies in the digital era, driving cybersecurity commitment, protecting personal data and delivering global alignment for stronger cyber resilience. 

Rosamund Powell, research associate at The Alan Turing Institute, explains that AI systems have multiple cyber security risks attached, which often go unrecognised as developers focus on introducing new services. Powell believes the new measures provide practical support to developers in creating a secure approach within their AI design and development process.

Plans to embed this as a global standard are vital based on the central role international standards play in managing AI safety challenges. Studies indicate the need for an inclusive and diverse working group, supported by incentives and upskilling, ensuring continued progress and success of these global standards.

‍