Educating French decision-makers on AI existential risk

Paris, France

A dedicated role at Pause IA to brief French policymakers on existential and catastrophic risks from advanced AI, and to train our volunteer network to do the same with their own representatives.

Endorsed by+1

A dedicated role at Pause IA to brief French policymakers on existential and catastrophic risks from advanced AI, and to train our volunteer network to do the same with their own representatives.

Endorsed by+1

Project Details

We are the French branch of PauseAI, and the only grassroots movement in France focused on catastrophic risk from AI. Since our launch in mid-2024, we have achieved a lot with very little funding:

Over 1.6M views from long-form YouTube interviews, plus coverage in major press
A conference at the French Senate on AI risks (October 2025), and the largest French-speaking AI safety conference to date (february 2025)
11 local groups across France, around 30 active volunteers, ~1,400 people on our Discord

We ran on volunteer work alone, with very little money, until late 2025: we spent around €21K in total up to December 2025 for the whole organisation, which is almost nothing. That is the main reason we think we could do much more with more funding.

Today, almost no one is explaining AI extinction and catastrophic risk to French decision-makers and politicians. As a result, the people who actually make the decisions are largely unaware of the problem. We already have the network and some credibility, and we try to do this work with our small team, but nobody has the time to do it.

Our only employee is already fully saturated and our volunteers already have a lot to do with limited time available.

The bottleneck is not the ideas or the network. It is having someone with the time and capacity to actually do the work.

The new hire would have two linked roles.

First, doing the work directly:

Meeting MPs, senators, parliamentary staff, and other influential people in politics and decision-making, to inform them about AI extinction and catastrophic risk
Producing the material these exchanges need: arguments, briefs, answers to common objections

Second :

training and equipping our volunteers and our community so they can raise the subject with their own representatives across France.
Our existing community is an effective lever here: they already care about catastrophic and extinction risk from AI, they are spread across the country, and elected representatives are more open to a conversation when it comes from their own constituents.

Over the funded period (half a person-year), we would expect to:

secure 30 to 40 meetings with MPs, senators, parliamentary staff, or party and campaign teams.
build a reusable set of materials: briefing notes, argument sheets and a practical guide for the volunteers on how to talk to their representative about AI risk
train 15 to 20 volunteers to raise the subject with their own elected representatives
have several local groups actually engage their representatives (we already use our locals groups to engage with representatives)

What we expect to deliver is network of relationships, private support from some of the people we meet, and a volunteer network that can keep the conversation going.

On a half-time role we are not promising public commitments from major political figures within the period (that may take longer).

The role would be managed by our Executive Director, Clémence Peyrot, currently our only employee. We already have a recruitment process (the one we used to hire her in December 2025). Nobody is pre-identified for the role; recruitment would start on funding.

We are strictly non-partisan. We do not support any party or faction; our goal is to educate and inform.

Theory of Impact

At this point, decision-makers are probably the most effective level at which to act to reduce x-risk. AI labs will not slow down on their own, and will not put serious effort into alignment unless governments force them to through regulation. Governments are also the only place where international coordination, or a slowdown or safety treaty, can realistically come from.

France is one of the most influential countries on European regulation, and so far it has taken a pro-business, pro-acceleration position: it worked to weaken the European AI Act, and it turned what should have been an AI Safety Summit into an "AI Action Summit". So moving France is not a side issue. It is probably one of the few European tipping points.

Today, French decision-makers are very badly informed, and almost nobody is trying to close that gap. The debate in France is dominated by voices that minimise or outright reject AI extinction risk. Many elected officials and their staff have limited time to assess technical arguments, encounter conflicting claims about advanced AI, and may not have access to concise resources that distinguish current harms from more severe loss-of-control scenarios. We have already done some work to change that: our president debated Luc Julia, probably the second most influential AI voice in France after Yann LeCun, and someone who shares LeCun's dismissal of AI x-risk. The debate prompted several press articles and youtube videos questioning his expertise, and he is now visibly less cited as an authority on the subject. In short, almost nobody is explaining catastrophic risk to elected representatives, which makes this a particularly neglected area, and neglected areas are where a marginal dollar goes furthest.

People

Clémence Peyrot

Executive Director

Maxime Fournes

President of Pause IA

Grants Received

Add grant

Communications Director role (2026)

from Founders Pledge

Future of Life Institute grant to PauseAI Global (funded our ED salary for some time)

from Future of Life Institutefutureoflife.org

Seed grant from PauseAI Global (2024)

from Stichting PauseAI (PauseAI Global)

$11,000

Funding Asks

grantmaking.ai Launch Round

Applied

Minimum

$36,000

Ideal

$72,000

How the money will be spent

Entirely on the loaded employer cost (gross salary plus French employer charges) of one person doing this work. No overheads; our operating costs are already covered.

$36,000 buys half a person-year, and how we use it depends on what else we secure for the same role:

If this grant is our only funding for it, we would likely run it as 12 months at 0.5 FTE. Advocacy runs on relationships with policymakers, and those take time to build, so a longer steady presence is probably more effective than a short push.
If we also secure the complementary funding we are seeking elsewhere for this role, we would combine them into a full-time position (your grant covering roughly half of it), which is the stronger version of the work.

Either way your money funds the same thing: a person doing this work, for half a person-year.

The ideal ($72,000) would fund the role full-time for a year, which is the version we actually think this work needs.

The $36,000 is the minimum for the role to be effective because finding the right person and getting them up to speed on both the technical substance and our network takes real effort and the equivalent of six months of work is the minimum period to achieve results. We would rather not run the project than run it badly with less than the minimal amount.

Discussion

Private comments in this thread are only shown to approved funders and grant reviewers. Email hi@grantmaking.ai or to request access.

Pip FowerakerEndorsement11d

I buy the same ToC and impact arguments as for other Pause AI chapters.

Romain Deléglise11d

Thanks!

Private UserPrivate11d?