Something curious happened in Silicon Valley last week. Three of the wealthiest, most powerful men on the planet — the CEOs of OpenAI, Anthropic, and SpaceX — gathered in chorus to warn the world that artificial intelligence might kill us all. They urged a slowdown. They spoke of existential thresholds, of superintelligent machines spiraling beyond human comprehension, of apocalyptic risk. The financial press ran it breathlessly. The cable news cycle followed.
And while the cameras were pointed at the billionaires performing concern, the actual, measurable, ongoing harms of their technologies continued exactly as before: Palantir software tracking immigrants for ICE. Flock Security cameras logging license plates on nearly every arterial road in dozens of American cities. AI-assisted targeting coordinates fed into military strike systems. The surveillance architecture of the 21st century, quietly expanding into every neighborhood while its architects claimed to tremble before their own creation.
The Convenient Fog of Existential Risk
This is not a coincidence. The language of AI apocalypse is extraordinarily useful to the people profiting from AI’s present harms. When the dominant narrative frames the problem as a future extinction event — alien, abstract, incomprehensible — it crowds out conversation about the harm already here. It reframes billionaires as reluctant stewards of dangerous fire rather than engineers of a surveillance economy. It invites regulation designed to entrench the biggest players while blocking competition from smaller, community-controlled, or open-source alternatives.
As journalists at The Intercept and 404 Media have documented, the privatized surveillance apparatus has become a genuine wild west. Technologies once requiring NSA-level budgets and state intelligence apparatus can now be purchased off-the-shelf by a contractor in Virginia, a county sheriff’s department, or a corporate HR team. The warrant requirements that once created at least nominal friction have been effectively dissolved by routing surveillance through private intermediaries. The data exists. It flows. And it flows toward power.
But here is what the doomsday narrative would prefer you not notice: communities are fighting back. And the coalition assembling against privatized surveillance infrastructure is unlike anything the tech industry has faced before.
From the Deepest Red to the Room
The backlash against Flock Security — the license plate reader company whose cameras have proliferated across the country in deals with local police departments — has been described by investigators as something rare: genuinely nonpartisan. From libertarian conservatives to progressive abolitionists, from rural gun owners worried about government overreach to urban organizers tracking police targeting of Black neighborhoods, opposition to mass vehicle surveillance has found common ground across ideological lines that usually hold firm.
This cross-pollination matters. Movements that rely on a single ideological base are fragile — they can be isolated, labeled, dismissed. Movements that emerge from unlikely coalitions are structurally harder to contain. When the same city council meeting draws someone in a MAGA hat and someone from a prison abolition collective, both there to oppose the same Flock contract, it signals something deeper than partisanship: a shared intuition that permanent, ambient, frictionless surveillance is incompatible with freedom, full stop.
The tools of this resistance are unglamorous but effective. Public records requests. City council testimony. Local ordinances. Investigative journalism by small outlets operating outside the legacy media apparatus. These are the instruments of accountability when formal institutions fail — and they are working. Cities have voted down Flock contracts. Lawmakers have introduced bills requiring transparency about surveillance purchases. School districts have banned classroom AI monitoring tools.
Reclaiming the Machine
The deeper question is not whether AI can be made safe in some abstract, future-facing sense. It is who controls the infrastructure of intelligence in the present — and whether that control can be meaningfully distributed.
Open-source AI models, community mesh networks, privacy-first communication tools built outside the venture capital economy: these are not utopian fantasies. They are real, functional, and increasingly capable. The Signal protocol underpins encrypted communication for millions of people worldwide. Mastodon and decentralized social networks have demonstrated that federated infrastructure can resist centralized control. Researchers at universities and nonprofits are building image recognition tools specifically designed to detect surveillance cameras and flag them for community mapping.
None of these tools are silver bullets. They require maintenance, adoption, and communities of care around them. They require the same kind of collective commitment that mutual aid networks require — not a product launch but a practice. But they represent a genuine alternative architecture to the one Silicon Valley is building: surveillance capitalism’s endgame, in which everything is tracked, everything is scored, and accountability flows only upward toward those with the money to buy the data.
The Accountability the Doomsayers Refuse
When Sam Altman warns of AI extinction risk, he is, at minimum, asking us to hold him accountable for a hypothetical future catastrophe while he profits from the real one happening now. That is not a trade communities should accept.
The lesson from the cross-partisan resistance to surveillance infrastructure is that accountability does not have to wait for federal legislation or for tech billionaires to develop consciences. It can start at the city council level. It can start with a public records request. It can start with a room full of people from completely different political traditions who all agree that they do not want a camera tracking their car every time they drive to the grocery store.
That kind of organizing — concrete, local, stubborn, and coalition-based — is harder to co-opt than a single-issue campaign. It builds relationships across difference. It produces wins that compound. And it refuses to wait for the existential risk summit to conclude before demanding accountability for the harms already accumulating, block by block, database by database, in the world we actually live in.
The doomsday future is not inevitable. Neither is the surveillance present. But one of those requires organizing. And organizing is always already underway.
