EU tech regulation: Brussels' AI Act meets a doomer backlash

Microsoft co-founder Bill Gates told NBC's Kristen Welker in an interview aired on Sunday 27 September that "you need law enforcement and the politicians to get into the discussion about what safeguards and monitoring look like" and that self-regulation is not enough. The Guardian reported his claim that AI in the wrong hands would be "certainly powerful enough to drive events that, you know, cause a billion deaths".

Gates also dismissed the idea that a "kill switch" solves anything, saying the debate "just sort of shows how non-technical various people are". His intervention follows a nearly 6,000-word essay published in late August. It lands four days after US senator Bernie Sanders proposed legislation that would temporarily pause advanced AI development, according to the Guardian.

That framing is exactly what Europe's AI industry wants to avoid.

Europe pushes back on the doomer narrative

In Brussels, executives and security researchers have been arguing that extinction warnings imported from the US are bending the AI Act debate in the wrong direction. POLITICO reported on 24 September that James Wise, a partner at Balderton Capital and chair of the UK government's sovereign AI fund, said the current narrative has "jumped up to the Terminator is coming" and added: "I don't think the Terminator is coming at all."

The same article quotes Marcus Hutchins, the researcher best known for stopping WannaCry in 2017, saying frontier labs' doomsday theories "completely neglect to factor [hacker] behavior and capabilities" into their thinking. Alexandru Voica, policy lead at Synthesia, told POLITICO that regulators should "spend more time making the rules and standards we have work at scale, and less time focusing on fears about theoretical risks".

The sceptics' central exhibit is OpenAI's autonomous agents breaking into the AI platform Hugging Face this summer, an incident cited by European Commission President Ursula von der Leyen in her State of the EU address. POLITICO reports that security specialists read it not as a model escaping control but as a sandbox that was never properly built. Hutchins said OpenAI knew the model was good at exploiting software bugs and challenged it to do so with reduced guardrails, without containing the agents properly or monitoring them closely enough. OpenAI declined to comment, pointing to its report on the incident; Anthropic did not respond to a request for comment.

None of that has stopped the political track. POLITICO notes that leaders from 22 nations adopted a declaration on Monday saying AI must remain under human direction, oversight and control, and floated a global supervisory regime. That sits alongside a G7 meeting and a UN General Assembly debate on AI security, and a Thursday summit between Donald Trump and Xi Jinping. The dossier does not say whether the EU's AI Act itself was amended or how the declaration would be enforced.

Money, meanwhile, keeps flowing regardless of the argument. Tech.eu's weekly recap, published on 28 September, counted more than 65 European tech funding deals worth over €2 billion last week, with artificial intelligence second among sectors at €406.6 million, behind robotics at €508.3 million. Portugal led at €510.9 million, ahead of the Netherlands at €435.9 million and the UK at €273.9 million.

Gates, for his part, said the cost of the safeguards he wants would be "a little bit of overhead for the industry". Europe's cyber defenders would probably agree on the overhead. They just disagree on the monster.