Stanford, are you awake?

Outside this campus, something is happening that is too loud to ignore. AI’s development accelerates while democratic institutions erode. Wars are processed through language models. Tech workers are laid off in the name of AI automation. College graduates are competing with the AI hype for entry-level positions.

Our group was born to address this momentum. It was born from the shared conviction that the rapid development of AI must be met with equally rigorous critical inquiry into how these technologies reshape power, equity, and governance. It was born of serendipitously crossing paths with people across generations and disciplines, and realizing we shared the same cognitive dissonance: what was happening outside campus was too loud to be ignored in classrooms.

Stanford holds significant normative power to shape our collective technological and political reality. We did not find each other through a formal call; rather, we met through the shared discomfort and hope that our collective energy could be the basis of inquiry and change.

We call our manifesto a Manyfesto as it was crafted with multiple hands.¹ This is our opening statement. It is not the final word. It is an invitation to seize the momentum.

On the Moment We Are In

We write from Stanford, which is not a neutral location. There is no Silicon Valley without Stanford, and no Stanford without Silicon Valley. From lean start-ups to foundation models, the very grammar of disruption was codified on this campus. Stanford trades as much in narrative diffusion as in capital. It holds real epistemic power over how AI is understood, built, and adopted worldwide. Likewise, we recognize what Stanford embodies. Stanford’s own history renders its neutrality impossible. From dispossessed Indigenous land to eugenicist experimentation, power has largely flowed in one direction, towards racial, power, and wealth concentration. We believe it can be redirected. Critical AI exists to redirect Stanford’s authority toward public value, democratic accountability, and equitable AI governance.

We write from a pivotal time of political and social unrest. In the United States and in the world, democratic institutions are eroding. Polarized discourses fuel anti-democratic movements. Tech manifestos recently released from Silicon Valley intend to mainstream techno-fascist and white supremacist ideologies, justified on the basis of the AI arms race and imperialist claims. Again, we can course-correct. This is not the only trajectory. Other narratives exist within Stanford and Silicon Valley. Here and around the world, the tide is already turning. From AI resistance reading lists to tech workers unionizing to the Vatican questioning the neutrality of AI discourse, people with differing sets of values are defying the self-serving AI sales pitch. AI’s trajectory is not inevitable. Our responsibility is to question the stories that make it appear so. From Business to Law Schools, it is no longer credible to unquestionably drink the AI Kool-Aid.

What We Believe

1. AI is a socio-political technology. AI development does not emerge in a vacuum. It is co-produced by the institutions, labor arrangements, and political economies surrounding it. Treating AI as a purely technical phenomenon, a set of capabilities to be celebrated or feared in the abstract, mistakes its nature and disarms our response to it. Preventing AI harms and risks goes beyond technical intervention. While evaluations, red teaming, and benchmarks are useful tools, they can only be effective when targeting the right set of social issues and meeting mandatory regulatory compliance.

2. AI should be governed as a normal technology.² We believe AI should be treated like other general-purpose technologies, such as electricity, aviation, and the internet: subject to gradual diffusion, sector-specific regulation, and human control, rather than as a rogue superintelligent entity operating outside the reach of existing institutional frameworks.

3. AI’s trajectory is not predetermined. The dominant narrative presents AI’s development as a force of nature: inevitable, exponential, beyond democratic reach. We reject the grammar of inevitability. As Terry Winograd observed, deciding is a computational activity, something that can ultimately be programmed; choice is the product of judgment, not calculation. The difference between deciding and choosing lies between submitting to an algorithm and exercising human agency. We are here to defend that distinction, and to insist that the choices being made about AI, by firms, by states, by universities, are choices, not computations, and must answer to those they affect.

4. To be critical of AI is to be reasonable. Our inquiry is not guided by nostalgia, fear, or resentment but by the desire to scrutinize any powerful system. We ask who benefits and who bears the costs of this technology. We reject the charge of technophobia. Our critique is not a rejection of technology; it is a demand for accountability. Critical inquiry is necessary if AI is to serve public value rather than concentrate corporate power.

5. To be critical of AI is to question power and information asymmetries. It means asking the questions industry narratives often avoid: Whose interests does AI amplify? Whose futures does it foreclose? Whose labor does it conceal? And which alternatives does it render unthinkable? Critique helps us hold contradictions in view long enough to preserve nuance and still gain the clarity to act.

6. The most important AI harms are currently hidden. Mainstream AI discourse is being co-opted by the disproportionate funding from entities with corporate conflicts of interest, such as AI labs, big tech, and effective altruist groups. Attention must be given to these funding sources: how they normalize AI discourse, direct our attention, and shape policy agendas. Currently, disproportionate attention is given to speculative risks such as longtermism, superintelligence, and catastrophic threats, instead of AI’s material consequences: mass domestic and foreign surveillance, militarization, environmental degradation from data centers, harms to children and vulnerable users through AI sycophancy, and the exploitation of hidden labor across the Global Majority. These concrete harms are strategically buried and need to be resurfaced for intervention.

What We Are Not

We are not technophobic. We are not calling for a ban, a moratorium, or a return to analog. We believe critique alone is insufficient. We deconstruct in order to reconstruct.

We are not naive about power. This group exists within an institution deeply entangled with the AI industry. We do not claim to stand outside of it. We claim to name these ties to resist its gravitational pull and conformist thinking.

We are conscious of our own power. We are conscious that we are operating in an environment that did not expect us to emerge, but also where our critical thinking is perhaps most necessary.

What We Are Doing

The Stanford Critical AI Group has three pillars:

1) Community. We are building an interdisciplinary community of critical thinkers across campus, ranging from undergrads to faculty. We nurture an environment where difficult questions can be asked safely, with compassion and empathy. While our group is focused on Stanford affiliates, we welcome critical discussion and collaboration with like-minded groups beyond campus.

2) Research. We have launched working groups to advance scholarly AI research. Our groups are currently investigating the political economy of AI; surveillance pricing; the history and ideology of tech manifestos; and alternative AI regulatory frameworks that prioritize the governance of AI as a normal technology.

3) Advocacy. We are prototyping forms of resistance through workshops and convenings. We bring missing voices and debates to campus, and collectively re-write the language of resistance through our AI Solidarity Lexicon.

Our Invitation

This is a Manyfesto, written by “many” of us. It is an opening, a question, a dance. An invitation to shape our common trajectory. We invite scrutiny, revision, and disagreement. We embrace debate and discomfort, as it’s the only way to provoke change. What is not optional, at this stage, is critical engagement. We are at a pivotal time, where the choices being made now, about how AI is built, deployed, governed, and narrated, will shape the world for decades. Stanford plays a role and we have decided to step in.