An Anthropic employee quit with a stark warning that the artificial intelligence models being built "could kill us all by the end of the decade." | Riccardo Milani/Hans Lucas/AFP via Getty Images
BRUSSELS — A viral social media post warning that top American AI developers are "gambling with our lives" has been met this week in Brussels with business as usual — and a splash of Schadenfreude.
The European Union's corridors of power were one of the few places over the past 48 hours not overtaken by a frenzy after an Anthropic employee announced he was quitting, with a stark warning that the people building AI "earnestly believe" that the artificial intelligence models they're building "could kill us all by the end of the decade."
Multiple EU policymakers responded to questions about such a threat by noting that Brussels already has a law for that: the bloc's Artificial Intelligence Act.
Advertisement
Others noted that the dire AI extinction warning could actually be a chance for the bloc's highest officials to regain control over the AI narrative. Since the groundbreaking 2022 release of ChatGPT, the EU was repeatedly mocked by tech enthusiasts for regulating a still-emerging technology.
European officials say they can now claim they'd understood the risks that others are only now waking up to. "In my view, the EU has already legislated for exactly this scenario: it is called the AI Act," said Michael McNamara, an Irish European Parliament lawmaker, who's the co-lead of the Parliament's group monitoring AI.
A leading European AI CEO, Uljan Sharka of Italy-based Domyn, which develops models for regulated industries, called Tuesday's post on X by Jacob Coxon, the former Anthropic and OpenAI researcher: "ridiculous." Sharka argued that the panic set off presents a "massive opportunity" for European Commission President Ursula von der Leyen at her State of the European Union speech next week to "explain why the AI Act exists."
Michael McNamara, an Irish European Parliament lawmaker, who's the co-lead of the Parliament's group monitoring AI. | Gianluigi Guercia/AFP via Getty Images
The "keep calm" attitude has been the mantra among EU policymakers: don't get carried away by the hype or incidents; trust the law Europe has in place.
"The EU already has a legal framework to regulate and mitigate the risk posed by advanced models," said European Commission spokesperson Thomas Regnier in a statement shared with POLITICO.
Systemic risks
Under the bloc's AI law, companies like OpenAI and Anthropic must take into account and counter what are called "systemic risks." These include the risk of losing control of a model, or the risk that a model is no longer aligned with human goals and values (known as "misalignment").
Advertisement
The Commission's Artificial Intelligence Office enforces this monitoring, and has contracted leading independent AI researchers, such as those from METR, an independent non-profit research organization based in California, to carry out the regulatory oversight.
Indeed, it was METR that was called in by OpenAI researchers to investigate after a much-discussed July hack of AI company Hugging Face, when OpenAI-powered agents went rogue.
One sign that Brussels wants to focus on enforcing the current lawwas the confirmation Thursday that European Parliament lawmakers would raise the AI extinction warning for discussion on Sept. 30, as part of a regular enforcement check-in with the European Commission.
They urged the Commission to increase the pressure on the companies and to enforce the law.
"This is exactly why the AI Act created obligations for systemic-risk models. Europe must now enforce them fully," said Italian Social-Democrat Brando Benifei, the Parliament's other lead of the group in charge of monitoring AI.
Domyn's Sharka said that von der Leyen should use her State of the Union speech on Wednesday to lay out a vision for how Europe can compete in the AI development race without sacrificing its leadership in regulating the powerful technology.
"One of the points where Europe is failing is that it does things in the right way, but we're not explaining them enough," Sharka said. "This is not about slowing down innovation; this is about building actually even faster, but with a safe approach."