Andrew Ng told Bloomberg TV that warnings about AI wiping out humanity are ‘much more science fiction than science’. Ng co-founded Google Brain and built Coursera, and he worked alongside both Dario Amodei and Sam Altman earlier in their careers.
He didn’t stop at disagreeing. Ng says the industry stokes these fears deliberately — for attention, and to push regulation in a particular direction. He dates the current wave to 8 September, when an Anthropic researcher resigned and accused his former employer of gambling with people’s lives. That post reached 150 million views.
The ladder being pulled up
The accusation isn’t new from Ng. In December 2023 he told a US Senate forum that large companies inflate risk to get rules that ‘pull up the ladder’ behind them — keeping new entrants out. In the same statement he put the odds of AI causing human extinction at roughly one in ten million over a century.
Three years on, the wording has barely moved. What changed is the volume. We wrote yesterday about Jacob Coxon, whose resignation set the debate off; he says the response surprised him. Ng reads the same wave as a campaign.
Europe wrote both camps into the same law
This is where it gets interesting, and for that you have to look at Brussels. The EU AI Act lists systemic risks: lowered barriers to chemical, biological, radiological and nuclear weapons, cyber capability, disinformation, major accidents in critical sectors. That is precisely the list Ng thinks is the right one.
The same recital also covers loss of control — unintended issues with alignment to human intent, models making copies of themselves. Both sit in one paragraph. The Commission’s tech chief Henna Virkkunen said this month that EU law requires Anthropic and OpenAI to assess the risk of losing control of a model, and that this is not the case globally.
Whoever sets the threshold settles the argument
Those duties only bite above a compute threshold. They land on the largest developers and on nobody else.
Which is exactly the ladder Ng described in 2023 — except this time no lab built it, a legislature did. And it sits awkwardly with his thesis: if regulation burdens the big players and spares the small ones, it’s a poor instrument for shutting out competition.
That leaves his second claim, and that one is harder to test. Whether the warnings are sincere or public relations can’t be settled from outside. What can be checked are numbers — and Anthropic published some for the first time yesterday. That’s what this fight is really about: not motives, but the fact that until now nobody could do the arithmetic.
Sources: