You Should Come to The AI Protest

cart;horse: If you are in the Bay Area on July 11th, even if you're at a company being protested, you should come to The AI Protest. It's fully legal and nonviolent (we'll have a full overtime SFPD escort the entire time), and it's not the worst way you can use your Saturday afternoon that weekend. Plus all of your coolest friends will be there.

There's a lot of discussion about the effectiveness of protests and marches; I don't want to re-litigate that here when you can just ask your favorite model.^[1] There's also lots of existing discussion on if/how/when we should pause, I'll point you to Katja's recent post and the discussion there as a start for that. This is a review of some concerns I have had or still somewhat have about this type of action, and how I've reasoned about and gotten through them. I imagine you might also have these concerns.

The ask (stopping the race) is one that a large fraction of the public is likely supportive of, with the majority concerned about the current pace of development. Of course there is a ton of behind-the-scenes inside-baseball work to be done and the details do matter, but we do need to build the pressure and steer the conversation from the outside, so that those on the inside get the resources they need towards making it happen. The larger the demand that frontier AI companies, governments, and other actors put forth concrete proposals and start coordinating, the faster that it will happen.

Coalitions and Communication

There will be people who care about data center water usage. There will be people who explicitly never use AI. There will be people who are concerned about CSAM and child suicide.

But this protest is about a push for a conditional pause, specifically because of transformative AI leading to the value of human labor going to zero, gradual disempowerment and concentration of power, and catastrophic and existential risks.

Coalition building is a necessary part of dramatic and positive change, and this march is a first step toward that. This is a fantastic opportunity to understand and shape the conversations happening outside the EA/Rat/AI bubble; many people there won't be reading the EA Forum or LessWrong. The more they see the passion we bring to naming these risks, and that we're all fighting for the same values, the more likely we are to succeed.

A Few Rebuttals

"This will affect my future career prospects at [frontier company]." Realistically, your participation is unlikely to be used against you: a background check coming back "attended anti-AI protest" and costing you a job isn't a serious threat model (a potential exception being you have strong evidence your counterfactual impact depends on a specific job, and strong reason to believe that employer would actually reject you over this). There are enough people trying to play the inside game, and I think we need those on the outside to be playing their part more strongly (making their opinion louder). Plus you don't have to worry about value drift :).

"This looks like I'm protesting my own [frontier company]." The protest is specifically focused on Stopping the AI Race, and your company has gestured at this in principle but has stopped short of committing. You're not there in opposition to your employer; you're there in support of a shared goal. The hard work of operationalizing a slowdown/pause/stop still has to happen, but the more pressure exists to make it happen, the faster and more seriously it will be taken.

"I'll be photographed or identified, or I could face HR or reputational consequences." It's possible. There are protections for at-will employees under California law (Labor Code 1101/1102), but you might be under some other term that makes this effectively moot. There's nonzero risk here, but you can attend without being the story. There will be hundreds of people there and the news cycle will likely move on immediately. There's also power in numbers! Get coworkers and come conditionally: if ten of you come, it's now about "employees attending", not about you specifically.

"I'm at [frontier company], I'll be called a hypocrite." This is exactly the kind of credible signal that only people with knowledge "on the inside" can send. The nurses, artists, and climate advocates at this march don't have the credibility to say "I understand the capabilities; I'm building them; and I'm still asking you to slow down." You do. That's a HUGE signal to the broader public and to policymakers about what kinds of humans cared enough to show up, and their credibility. This is an easy way you can leverage your position to effect positive change on the outside, without having to actually leak anything.

"The crowd will be radical and I'll be associated with it." There probably will be people saying "Fuck OpenAI" and "Quit Your Job." There will be rough takes, and speeches or chants that come from a perspective you don't specifically endorse. This is a part of coalition building, and while group dynamics may come into play, most people will have individual perspectives that you can both learn from and help inform! Speaking to these people in person, as people, will make the nuance you and I care about come through more strongly. If you stay home to avoid the noise, the noise is all that's left.

Attending, in my opinion, is brave: you have values, you want to shape the world toward them, and you'll compromise with others and take on some personal risk for a shared cause. In the situation where everyone reasons the other way (i.e., if all of us stay home to protect our own standing), no one is left to say in public what they believe in private: a far worse outcome than being occasionally spotted at this protest.

This is a moment where your values, the frontier companies' stated values, and a broad public coalition are starting to point in roughly the same direction. This doesn't happen often, and your marginal attendance can help steer that rough direction. Please don't sit it out!

The march is happening on Saturday, July 11 from 12–3 PM, starting at the OpenAI headquarters through downtown to Anthropic and DeepMind, with speeches at all three. More details at:

https://theprotest.ai/

https://stoptherace.ai/

Luma - Partiful

I had a large list of clauses and bullet points, made rough paragraphs from them, had specialized Claude Opus 4.8 instantiations make suggestions, accepted a bunch of those, then read them back and didn't like them, and ultimately rewrote a bunch of it. I'd estimate 10% of the clauses in this post were generated fully by an LLM, the rest are fully mine with minor grammar/nit edits. I endorse every word of this post.

1. ^{^}

   Some first pointers synthesized from a first Claude pass (links verified):

   On the effectiveness of protest and its limits, see Chenoweth & Stephan, Why Civil Resistance Works, whose 1900–2006 dataset finds nonviolent campaigns roughly twice as likely to succeed as violent ones (and the source of the widely cited "3.5% rule"); Gillion, The Loud Minority, which treats protest as a signal that measurably moves legislative votes, voter turnout, and elections; Gause, The Advantage of Disadvantage, arguing that costly protest by low-resource and marginalized groups draws more responsiveness from legislators, not less; and Giugni (ed.), How Social Movements Matter, a useful check on all of the above for how hard it is to show movements actually caused the outcomes credited to them. For the skeptical counterpoint, see Tufekci, Twitter and Tear Gas (freely available online), which argues that digital tools let movements assemble enormous marches quickly without building the organizational capacity that sustained older movements — so sheer size can mask underlying fragility.