Chatham House can’t have known that their event “AI and National Security: Who’s Really in Control?”
would take place as Anthropic both launched Claude Mythos and attended talks at the White House to resolve their dispute with the US Government, but it does show their foresight that this is the crucial topic in defence at the moment. As panellist General Sir Richard Barrons, a Senior Consulting Fellow at the Royal Institute of International Affairs put it:
“AI unleashed will destroy civilisations“.
How then, the event considered, should AI be leashed and to whom?
Lawyers in the (metaphorical) battle-space
For our clients and for us as lawyers, the question is existential in different senses. How should government contract with the private sector, and the private sector within itself, in these extraordinary times and with this extraordinary technology, to achieve commercial certainty and preserve national security? Get it wrong and damages will be the least important measure of loss.
On the night, much was made of the power of the new technology. The General cited the US 1000+ missions a day in Iran, the “digital kill-map” in action. AI selects targets and deploys crewed, uncrewed and autonomous systems – the last with the power to bomb and shoot without human intervention. As well as its application at the sharp end, AI will organise logistics, such as supply and casualty evacuation. He warned of a military establishment, conservative in its views without the mindset to adopt to the new way of fighting in principle, let alone in practice as billions and billions of dollars worth of investment transform the technology at an almost literally unbelievable pace. And even if ‘our’ soldiers, sailors and airmen get on top of this new kit, “the enemy gets a vote”. All this is being deployed in an oppositional environment where ‘they’ will use ‘their’ AI to destroy ‘ours’ and ‘us’: and of course vice-versa: a digital arms race on an unprecedented scale.
Quis custodiet ipsos milites?
His faith is in humanity, in statesmanship to pull us and our systems back from the brink. Query is any other model of governance possible? John Thornhill, Innovation Editor at the FT, and recently back from Silicon Valley set out that in US tech circles there is no ‘debate’ as such but a multiplicity of views: few believe that AI companies should be a “handmaiden of the state” but just how many competing approaches to governance and ethical use questions can the eco-system sustain? Dr Yasmin Afina, a Researcher in Security and Technology in UNIDIR’s Security and Technology Programme, talked of the work being done outside the EU/UK/US, where government drives much more of the R&D so query the extent of role of the private sector in governance. At the same time, these governments are looking for reliable commercial partners whose delivery against political constraints will be certain and predictable.
Thornhill referenced the so called ‘6th Domain’: the first five are land, sea, air, space and cyber; and now there is the interface between private and public sectors in the field of AI and tech innovation. It is Ukraine’s success in this last that is enabling it to launch drone strikes 500 miles into Russia and why it is becoming a prime partner in contracting with European and Gulf states to augment their security in aspects of drone warfare.
How do commercial lawyers adopt to all of this? We should not look to nation states to impose any sort of effective governance. Article 22 of the London Naval Treaty of 1930, purported to prevent submarines sinking merchant vessels without first securing the safety of passengers and crew. Germany acceded in 1936. Doenitz was put on trial at Nuremburg for breaking this rule only for Allied sub-mariners to warn they would appear as witnesses for the defence, so roundly ignored was the law in practice. Any treaty on AI will surely go the same way. It’s been said in relation to this tech there are thousands of Oppenheimers, only now without two superpowers with a mutual interest in a terrifying status-quo to keep the World from burning.
Docusign to delivery
Our role is therefore to help find the balance between the state’s interest in control and the suppliers in both sales and development. Those interests are not always in competition: it is in the interests of Ukraine, for example, that it is seen as a dependable partner who to the extent it restricts use of its technology does so on a rational and consistent basis. Lawyers may help government and industry to produce and work to a framework – which must in turn be rapidly adaptable to keep pace with the tech – that means that agreements can be rapidly concluded. AI must be deployed in our nation’s defence as soon as practicable: lawyers can and should step up to help make that happen.
Liz Williams (Defence) and Jonathan Chamberlain (AI and Governance) attended the Chatham House event on behalf of Gowling WLG.
