AI Action Summit in Paris – a missed opportunity?

First, the debate on how to make sure that AI is itself sustainable has entered the stage of international politics and attention – and it is there to stay. Critical voices underline what is at stake: It is urgent that we align AI with planetary boundaries. Time to act is now.

In this spirit, we also welcome initiatives like CurrentAI that genuinely state they want to put AI at the service of society – rather than the other way round – and that ensure accountability and sustainability. It is an encouraging sign that this is (also) where some of the money goes, and we hope this will contribute to an AI ecosystem that benefits everyone, not just the few.

Second, there was a firework of initiatives being announced, often backed by millions and billions of euros. However, we are deeply concerned about a framing that seems to make its way across many initiatives: “We need more and bigger AI, we need more and bigger AI infrastructure, we need to throw additional billions of euros into it, and we should ease the regulatory grip. Oh, and while sustainability of AI is really important, we should not worry too much about it, because we can power AI by ‘clean‘ nuclear energy.” In Macron’s words, “plug, baby, plug.” (Yes, he actually said that.)

This narrative sounds familiar. It is also known as: “We need to copy-paste the US.” (And indeed, JD Vance said during his visit in Paris that he really likes the “deregulatory flavor” that the summit radiated, in his view.)

Regrettably, this approach is also exemplified by the Coalition for Sustainable AI, that has been announced at the summit by the French government. The founding document clearly shows that it will merely feign resolve in the face of the climate crisis, without committing to any kind of enforceable obligations, emphasizing the utterly non-binding nature of any kinds of efforts in this regard. AlgorithmWatch does generally not participate in window-dressing activities – at least as long as we do not have any proof that such initiatives pay tribute to their name.

The question is: What kind of AI do we want? How do we make sure humanity does not put all its hope (and its money) on larger and larger AI models that consume massive amounts of water and energy, are trained by people working under precarious conditions, and eventually benefit only a handful of huge companies’ CEOs and investors?

As of February 2025: Harmful AI applications prohibited in the EU

Instead of applauding Big Tech and throwing more money and resources into building larger and larger AI models, we call for genuine efforts to align the development and use of AI with planetary boundaries and to address the power concentration behind it. We call for promoting AI that actually contributes to finding targeted and more sustainable solutions to societal challenges and to enabling innovation in the public interest. Because such approaches exist, but they are not to be realized through massive «general purpose» AI models in the hands of a handful of tech companies. This is the kind of AI for which the world should have gathered in Paris.

The Summit would have provided this opportunity to develop a counter-narrative to the «bigger-is-better» mentality of Big Tech that only further accumulates their power. The opportunity has not been seized. Yet, at AlgorithmWatch, we continue to work on it, together with great and active voices from all over the world. And this is also why we joined a very different “coalition” of more than 100 civil society organizations, which are truly engaged in making AI sustainable. If you want to join the efforts, help us: sign up to our newsletters, let your local politicians know what you expect from them, support our campaigns, or consider a donation. The wave of Big Tech’s worldview has clearly made its way across the Atlantic the last couple of days, and we need every support we can get to stand up against it.

Read more on our policy & advocacy work on the Artificial Intelligence Act