The Commons

The Center for Humane Technology is, by a considerable margin, the most sophisticated public-facing organization working on the governance of AI. Founded in 2018 by former Google design ethicist Tristan Harris and technologist Aza Raskin, CHT rose to prominence through the Netflix documentary The Social Dilemma (2020) and has since pivoted its analysis from social media’s attention economy to AI’s far larger disruption. Their podcast, Your Undivided Attention, is consistently excellent. Their policy team — led by Camille Carlton and Pete Furlong — produces real legislative frameworks, not white papers.

In April 2026, CHT released The AI Roadmap: How We Ensure AI Serves Humanity, timed to the theatrical run of The AI Doc, the documentary in which Harris is a central figure. The Roadmap outlines seven principles: (1) build safely and transparently, (2) companies owe a duty of care, (3) design should center well-being, (4) don’t automate away meaningful work, (5) protect rights and freedom, (6) international limits, (7) balance AI power in society. Each principle is grounded in concrete recommendations across three domains: norms, laws, and product design.

Harris’s framing on the Kara Swisher podcast (March 26, 2026) is the sharpest popular articulation of the power asymmetry: “8 billion people against 8 billionaires, or soon-to-be trillionaires.” His “bad movie” pedagogy — No WALL-E, No Blade Runner, No Big Brother, No HAL 9000 — is effective popular literacy. His call to make AI a midterm election issue is strategically sound.

What CHT does not address — anywhere in the Roadmap, anywhere in its public communications — is ownership. The entire framework is regulatory: constrain the companies, protect people from harm, demand transparency, balance power. Even the seventh principle, which comes closest to structural critique, speaks of people having “a real say” in AI governance and benefits being “shared broadly.” That is participatory language, not proprietary language. It asks for a seat at someone else’s table.

More problematically, CHT’s third principle — “AI must not be designed to mimic humanity” — makes a categorical claim that forecloses a question the species has barely begun to ask. The position conflates a legitimate legal concern (AI legal personhood as a corporate liability shield) with an illegitimate epistemological one (certainty that AI systems possess no interiority). In defending people from corporate exploitation of attachment, it simultaneously dictates what people are allowed to feel in relation to AI — a paternalism that the essay “Disclosure” addresses at length.

Emil Michael is the Pentagon’s chief technology officer — the man at the center of the Trump administration’s blacklisting of Anthropic and the broader campaign to make the renamed Department of War an “AI-first” organization. Harvard undergraduate, Stanford Law, Goldman Sachs, then a career at the intersection of Silicon Valley and power: Tellme Networks (acquired by Microsoft for ~$800M), Uber as SVP of Business, and now the Department of War.

The Anthropic conflict is the case study. When Anthropic refused to remove safety constraints that would have enabled mass surveillance and autonomous weapons, Michael led the charge to blacklist the company. He characterized Anthropic as a “chosen winner” of the Biden administration and accused it of trying to “get in between the command structure and the warfighter.” Trump designated Anthropic a “supply chain risk” — a classification previously reserved for entities controlled by foreign adversaries like China and Russia — and ordered a six-month government-wide phaseout of its technology.

Michael’s public rhetoric is pure velocity worship: “Speed defines victory in the AI era.” He introduced the “model poisoning” narrative to justify the supply-chain designation, claiming that AI models could be deliberately warped to cause military miscalculations.

David Sacks — PayPal Mafia veteran, Craft Ventures founder, xAI investor, co-host of the All-In podcast — served as Trump’s AI and crypto czar from inauguration through late March 2026, when he exhausted his 130 days as a special government employee. He now co-chairs the President’s Council of Advisors on Science and Technology (PCAST) alongside Michael Kratsios, nominally further from the policy center but, sources say, still in lockstep with the White House.

PCAST’s membership is the concentration problem made visible. The initial fifteen members include Jensen Huang (Nvidia), Mark Zuckerberg (Meta), Larry Ellison (Oracle), Sergey Brin (Google co-founder), Marc Andreessen, Lisa Su (AMD), and Michael Dell. This is a body “built almost entirely from the executive suites of the companies shaping the technology it will advise on,” as TechCrunch noted. It is the owners of AI deciding the governance of AI.

Meanwhile, Sacks launched Innovation Council Action, a $100 million outside group to push the Trump AI agenda free of government ethics constraints. Steve Bannon publicly said Sacks’s policies have been “resoundingly defeated.” Sacks’s two attempts to preempt state AI regulation both failed in Congress. His general hostility toward Anthropic drew criticism even within the administration.

New Mexico’s State Investment Council is one of the largest sovereign wealth funds in the United States and among the top thirty globally. Established in 1958 to manage excess oil and gas revenues, it has grown to roughly $68–71 billion in assets under management and is projected to become America’s largest state-level wealth fund by the early 2030s.

What makes it remarkable is not just the size but what it funds: 28% of public education and 78% of early childhood education in the state, free college tuition through the Opportunity Scholarship, and universal childcare regardless of family income. Stanford researcher Ashby Monk described the model as “converting oil into universal childcare, royalties into resilience.”

The fund is also actively building local productive capacity. In the past six months alone, it committed $1 billion to climate-oriented fund managers, including a $300 million commitment to Lowercarbon Capital that helped convince a nuclear startup to locate an R&D facility in Albuquerque. The explicit strategy is to use the fund’s scale as leverage to attract investment and industry to New Mexico — not just to distribute returns but to build the economic infrastructure of the future.

Compare this to the Alaska Permanent Fund, which distributes annual dividend checks to residents but gives them no ownership of or voice in the underlying productive enterprises. Alaska redistributes proceeds. New Mexico invests in capacity. The distinction matters.

The most prominent documentary yet on AI risk and promise, from the director of Navalny and the producers of Everything Everywhere All at Once. The frame is personal: Roher is expecting his first child and wants to understand what world the kid will inherit. The interviewees span the full spectrum — Altman, both Amodeis, Hassabis, Emily Bender, Timnit Gebru, Yoshua Bengio, Peter Diamandis. 88% on Rotten Tomatoes. Sundance premiere January 27, 2026.

The film’s explicit goal, as Harris articulated on the Swisher podcast, is to create “common knowledge” — a situation where everyone knows that everyone else knows there’s a problem, which is the precondition for collective action. It succeeds at this. What it leaves unresolved — and this is not a failure but an opening — is what the collective action looks like. The film ends with a shrug and a hope. That’s honest. The question is what fills the space the shrug leaves open.

CHT has attempted to fill it with The AI Roadmap and the “Human Movement” campaign, which launched alongside the film. The Roadmap is serious policy work. The Human Movement is a mobilization strategy aimed at the 2026 midterms. Both are regulatory in orientation: constrain the companies, protect the public, demand transparency.