Elizabeth Williams Partner
Gowling WLG
Michael Carter Partner
Gowling WLG
AI can unlock huge opportunities for defence. However, this is a large, disparate sector, so its safe, careful and targeted deployment is crucial.
Every industry must deploy AI safely, securely and in an operationally trustworthy manner. But for defence companies, this approach is critical. The repercussions of failure could be catastrophic for military personnel and national security.
Yet while AI implementation presents major challenges, responsible and considered use of the technology has the power to unlock huge opportunities.
“AI is used widely across defence platforms — and it’s going to be used more,” says Elizabeth Williams, Partner and Head of Aviation, Aerospace and Defence sector, Gowling WLG. “This technology can help us and our allies retain our military capability and edge more readily, more easily and at pace. If we don’t use it, we’ll be behind.”
No ‘one-size-fits-all’ approach
The Government takes a decentralised approach to AI, allowing each sector to apply its own guidelines. However, a major challenge for defence is its size, complexity and disparate nature. There can never be a ‘one-size-fits-all’ answer to AI deployment.
For instance, at the military end of the spectrum, AI is a central component of the Global Combat Air Programme (GCAP), a co-project between the UK, Italy and Japan to develop a next-generation fighter aircraft. Meanwhile, at the civil end, AI is being used to execute a wide variety of tasks, from product design to the streamlining of logistics.
Indeed, a recent report from IBM — called ‘Defense manufacturing in the AI era’ — predicts that by 2028, approximately 80% of manufacturing and supply chain operations will deploy agentic or autonomous systems for forecasting, inventory optimisation, scheduling, quality control and predictive maintenance.
“When considering legal and regulatory assurance, it is important to consider each use case separately on its merits,” agrees Michael Carter, Partner and Co-head of Gowling WLG’s Global Tech Sector. “The Government’s AI Practitioner’s Handbook (AIPH) states that ‘AI assurance activities can vary depending on the type of AI technology and how you plan to use it.’ So, it’s important to look at each AI tool individually and understand where it sits within the defence ecosystem.”
How can AI be deployed sensibly, carefully and securely with the right guardrails in place?
Supporting AI with improved procurement and contracting
While the sector has made great strides in facilitating AI adoption, improvements could be made in procurement and contracting. “This is technology that’s changing quickly,” says Elizabeth. “So, contracts need to be flexible to enable updates, continuous improvement and the ability for companies to operate at pace.”
She also notes that important AI innovations aren’t only created by government or large defence contractors. These are increasingly coming from smaller, fast-moving AI companies. To help SMEs develop their technologies and capabilities, systems must support easier collaboration, procurement and supply chain integration.
In 2021, the MOD set up an AI Ethics Advisory Panel to address these issues. “The Panel has been looking at this for a long time, so they’re probably further ahead on the curve than many,” says Michael. “So, with Government guidelines and principles, and the AI Practitioner’s Handbook, the sector is doing a lot to reduce friction for suppliers while maintaining high standards.”
The sector must still tackle questions regarding sovereignty, trust and values, and establish a position on AI providers active in other countries. “It’s balancing the opportunity and the increased capability offered by AI with factors such as security and authority data,” says Elizabeth. “There’s so much to unpack.”
But ultimately, the most important question for Elizabeth and Michael is: ‘How can AI be deployed sensibly, carefully and securely with the right guardrails in place?’ A key safeguard is always having “meaningful human control” in the loop.
