It is one thing for the computer scientists to warn that artificial intelligence could wipe out humanity in some science fiction-style cataclysm – an event that, while plausible, still feels rather distant.
But when Sir Ken McCallum, the director general of MI5, warns that AI is “already supercharging the cybersecurity contest between attackers and defenders” and is producing “tailored” propaganda to radicalise individuals, the existential threat feels much more tangible, if not imminent.
AI adds a whole new dimension to efforts by hostile states and terror groups to foment disorder, influence elections and divide societies. MI5 has, according to Sir Ken, “caught terrorists seeking to exploit AI to improve their reconnaissance and attack methodologies, with AI-enabled drones adding a new twist.”
However, in what he describes as this “new era” for counterintelligence, Sir Ken attaches the greatest weight to “AI’s potential to change the stakes on existing, or even new, bioweapons”.
As with the much-feared homemade “dirty bombs” using rogue nuclear material in a previous age, so now AI is capable of teaching even amateurish terrorists, let alone hostile nation states, to spread potentially devastating diseases in the name of some political cause or, more likely, use the threat of it to force governments to their will. A global plague created by AI is how human life on Earth could expire.
These risks are indeed very real and may not be far away. The variety of threats facing the UK has certainly never been more diverse, from domestic militants on the left and right, pursuing political objectives to the agents, formal or otherwise, of Russia, China and, as we have seen most recently, Iran.
Some of these international challenges can be almost blatant, such as the particular institute identified by MI5 presenting itself as a conventional research body, but in reality regularly used by Chinese intelligence to run projects in the UK on AI and cybersecurity – and directly using British taxpayer-funded facilities to boost China’s own espionage capabilities.
The secret intelligence services have dealt with everything from an arson attempt on Sir Keir Starmer’s home to sabotage of deep-sea communications cables, to an armed attack on a synagogue, to bots spreading lies about vaccine safety, to the devastating (Russian) cyber attack on Jaguar Land Rover, and the apparent plot to destroy US bombers at RAF Fairford.
It is an awesome array of hazards. But because AI is the first foe to be smarter than MI5, so to speak, the British intelligence agencies have to be able to use AI to counter the threat, and attempt to stay at least one step ahead. Wary of straying too far into political debate, Sir Ken and MI5 don’t promote policy solutions to the multifarious threats we face.
Nonetheless, some almost present themselves. As a powerful global phenomenon, and one with vast destructive potential, AI has to be regulated on an international basis. Britain and MI5 have the great advantage of access to some of the world’s most impressive AI research, and a world-leading regulator, the AI Security Institute. It is, thus, in a good position to provide leadership in this new field.
China and many Western AI chiefs have recognised the need for such global action to manage the rise of a set of super-intelligent entities that can out-think humanity. At the moment, though, President Trump’s preference for “self-regulation” – and a safety strategy based on blowing up data centres if needs be – is obviously unequal to the task of taming AI and keeping it within human command.
At some point, it can only be hoped, Washington will see the wisdom of applying the same framework of treaties and international agencies that restrict nuclear armaments to this equally lethal new technology. Under such a regime, MI5 and its counterparts around the world could concentrate, even cooperate, on isolating and neutralising the “bad actors”.
One smaller but hopeful development in the UK is the newly established National Centre for Information Defence, with a remit to “detect, attribute and disrupt” information attacks by foreign powers, many AI-powered. Whether the 2016 Brexit referendum was compromised or not is a critical question. But it is widely accepted that democracy today is more vulnerable than ever to destructive foreign interference – an unfortunate competitive disadvantage that dictators don’t face.
The same emphasis on preventative measures goes for all the efforts to build resilience across the national infrastructure – energy, transport, financial, defence, communications and much else.
All of this needs funding, and Sir Ken’s thoughtful and measured speech is a reminder to the chancellor, John Healey, also a former defence secretary, that there are few areas of public spending that are a wiser investment than the secret intelligence services. There is much for them to do, and they need the tools, AI included, to get on with the job.
Read MoreIf only Russia’s online bots had actually disappeared
Why aren’t our schoolchildren as revolting as the French?
The ghost of Rachel Reeves has delivered one last boost for Britain’s economy
Is AI really going to kill us in the next 40 years?
Andy Burnham is the PM who can overturn Brexit
Trump’s Middle East war has spiralled beyond America’s control
