Essay by Eric Worrall
A tangled web of murky motives, politics, underhand tactics and unexpected twists. But what else would you expect when it comes to AI?
First, I think we all owe a vote of thanks to Sabine Hossenfelder for exposing this mess. I don’t agree with Sabine’s climate views, but she has always struck me as a person of integrity, and this latest disclosure reinforces my good opinion of Sabine.
One of the organisations named by Sabine was “The Center for AI Safety”. Strangely according to Wikipedia “The Center for AI Safety” enjoys a lot of support from Open AI and other AI hyperscalers, or at least they did so in 2023.
Sabine also mentioned pro-AI groups are also promoting sponsored pro-AI content.
Sooo – we have an anti-Ai group which is or was until recently supported by US AI hyperscalers, and pro-AI groups which are supported by who knows who.
What on earth is going on?
So I did the obvious – I asked AI to figure out what is happening.
I just watched a video from Sabine Hossenfelder claiming she was offered money to tell us AI will kill us. Apparently there are also AI groups trying to pay influencers to produce pro-AI material. What makes it really odd is one of the anti-AI groups named by Sabine, “The Center for AI Safety”, appears to enjoy support from US AI hyperscalers.
Have you got any idea what is happening, who these groups are, and what they are trying to achieve?
This is real, recent, and genuinely tangled. Sabine’s video went up on September 5, and her core claim is that there is money in telling you AI will destroy humanity, and money in telling you it will improve your life — both sides are quietly paying creators, some with proper disclosure and some very much without. Here’s roughly who the players are.
The “AI doom” funding side. This money mostly traces back to the effective altruism / longtermist philanthropic network, above all Open Philanthropy (bankrolled largely by Facebook co-founder Dustin Moskovitz’s fortune). The Future of Life Institute and the Center for AI Safety share Open Philanthropy as a major donor, and the people in this movement — who tend to call themselves longtermists or effective altruists — have largely shifted their focus toward AI safety. This ecosystem now funds content creators directly: for example, there’s a “Frame Fellowship” that brings promising AI-safety content creators to San Francisco and supports them with mentorship, production, and funding so they can make content on AI risk. Critics like Perry Metzger argue most of the money comes from the EA movement, which spends heavily on shaping public conversation. Their goal, charitably stated: raise public alarm enough to get serious regulation of frontier AI. Uncharitably stated: manufacture panic.
The “pro-AI” funding side. This is newer, bigger, and more nakedly political. A super PAC called Leading the Future raised $125 million in 2025, backed by Andreessen Horowitz, OpenAI co-founder Greg Brockman, Palantir co-founder Joe Lonsdale, Ron Conway, and Perplexity, with the goal of backing candidates who support national AI regulations rather than state-by-state rules. It runs a network including the Republican-oriented American Mission PAC, the Democratic-oriented Think Big PAC, and an advocacy arm called Build American AI. The influencer angle is documented: a WIRED investigation found that Build American AI, tied to the super PAC, was paying TikTokers up to $5,000 per video to push anti-China, pro-AI talking points — with instructions not to tag or disclose who was paying. Their goal is concrete: kill state-level AI regulation, secure a light-touch federal framework, and win the 2026 midterms, where AI is shaping up to be a key issue and industry groups are spending heavily to counter public concerns about data centers, energy use, and job displacement.
Now the paradox you spotted — why does CAIS look industry-supported? A few threads untangle it:
First, the entanglement is genuine but runs through philanthropy, not the hyperscalers’ corporate treasuries. Open Philanthropy, the common donor to CAIS and FLI, also donated $30 million to OpenAI, and its former CEO Holden Karnofsky was once an OpenAI board member. The same Big Tech-adjacent wealth and social network that seeded the frontier labs also seeded the x-risk nonprofits. (Full disclosure worth making here: Anthropic, which made me, sits inside this same ecosystem — early Anthropic investors and staff have EA ties too. So read me with that in mind.)
Second, industry leaders publicly endorsed CAIS’s message. Its famous 2023 one-sentence extinction-risk statement was signed by the CEOs of OpenAI, DeepMind, and Anthropic. That’s what makes people suspicious: the cynical reading is that “our product might destroy the world” is fantastic marketing (it implies the product is godlike), attracts talent and capital, and invites regulation that only well-lawyered incumbents can afford to comply with — a moat against startups and open-source.
Third, though, the two camps are now literally fighting each other. CAIS’s lobbying arm backed California’s SB 1047-style safety bills; a16z and OpenAI fought them, and Leading the Future has opposed candidates associated with state-level AI regulation, such as New York Assemblyman Alex Bores over his AI safety bill. So “hyperscalers fund the doomers” is too simple — the industry’s political money is currently aimed at the safety camp.
The most defensible synthesis is probably this: there’s no single conspiracy, but there are two well-funded advocacy machines, both of which have discovered that YouTubers and TikTokers are cheaper per impression than lobbyists, and both of which blur disclosure rules. The doom side’s money comes from ideologically motivated philanthropy that happens to be socially entangled with the labs it criticizes; the boom side’s money comes from investors and executives with direct financial stakes. Sabine’s underlying point — that you should ask who paid for any AI opinion you encounter, including hers (her video is itself sponsored by Ground News) and including mine — is sound regardless of which side’s claims turn out to be right.
Source: https://claude.ai/share/37cf1b8a-6f30-4113-b949-47801641c400
Claude is run by people who are involved in this situation – so bear that in mind when you read Claude’s explanation.
The current battle might be as simple as Democrat vs Republican, a battle to influence the midterm elections. As WUWT has previously noted, some Democrats have recently pivoted to promoting AI fear, likely as a bipartisan replacement for the failed climate scare.
Whereas President Trump is promoting the position that AI is a critical US strategic priority.
Where or how China fits into this mess is anyone’s guess. Logic suggests China would like to retard the USA’s AI progress by stoking opposition to AI, but this whole issue is becoming so tangled where, how or even if China is playing a part is an unanswered question. For all we know, China is sitting back and watching the floor show, watching US pressure groups tear each other apart.
I guess the one takeaway we can be sure of from this mess is, be very careful who you trust on AI issues. I’ve personally never been offered money for any of my articles, probably because I’m not prominent enough to register on the radar of groups which make such offers, and I certainly wouldn’t be interested in accepting money which didn’t include a right of full disclosure, or which required me to promote a position I disagreed with. But there are a lot of pay to play influencers out there, and the infowar shadow game for control of the AI narrative is shaping up to be more tangled than the climate debate ever was.
