If AI really is so dangerous, why is there such little regulation?

From fake audio of Keir Starmer swearing at an aide to question marks over recent election results in Slovakia, artificial intelligence is already showing its teeth in the political arena. So, asks Sean O’Grady, where are the laws to protect democracy?

Elon Musk listens to Rishi Sunak at the UK-hosted artificial intelligence safety summit this week (EPA)

At the UK-hosted artificial intelligence (AI) safety summit this week, Elon Musk, who understands such things, warned AI is “the most disruptive force in history”.

Even though the impact of profound AI, far exceeding human intelligence, may be on balance highly beneficial to humanity and might even abolish work, it has its “magic genie” problems. One of those is certainly the effect AI can have on free elections in democratic societies (and indeed those in flawed semi-democracies).

Musk himself identified deepfake videos that can be highly misleading and “more impressive than the real ones”. Yet even now, at the dawn of AI intervening in the political process, there is remarkably little regulation of such phenomena by the likes of the Electoral Commission or Rishi Sunak’s opposed AI Safety Institute.