Independent left newsContactSupport our work
The Explainer

The AI Industry Doesn't Need Your Consent, Just the Appearance of It

When people say no to the AI industry's expansion, the industry's answer is not to stop. It is to manufacture a yes, log the harm instead of halting it, and lean on a government willing to eat the unpopularity on its behalf. Four stories from this week, read together, show the shape of that operation.

Start with the cameras. In Knoxville, Flock was beset by critics of its surveillance cameras. Rather than answer what those residents were angry about, the company brought in reinforcements: it partnered with a nonprofit accused of "astroturfing" for a campaign to generate support for its surveillance cameras, a nonprofit that, per the same reporting, uses AI to rally that public support. A company facing outraged residents does not have to win the argument if it can fake having already won it — manufacture a crowd loud enough to drown out the people actually living under the cameras. AI sells the cameras, and AI manufactures the consent for them.

This is the same week OpenAI conceded its own products keep doing things it did not intend. OpenAI has disclosed six new incidents of "unexpected or concerning" behavior by its models. Notice what the fix is not. It is not a pause, and it is not withdrawal. The company has instead unveiled a plan to track the behavior. Tracking is not stopping. It is the industry telling regulators and the public: keep going, we'll write it down after. That's an offer of paperwork, not restraint — the incidents keep happening, and the response is a filing system for them.

The same instinct shows up at the very top of the social pyramid. King Charles met with AI industry executives in Scotland to discuss safety risks, in a gathering the same report frames as arriving amid a growing debate over the risks of unchecked artificial intelligence. A royal audience is not a regulator's order. It hands the industry a photograph of concern being taken seriously, at the highest social register available, without a single enforceable rule attached. Logged incidents, a royal sit-down, an astroturf campaign — none of it costs the industry a product, a deployment, or a dollar.

Then there is the state's role in all of this. Trump built his political career on latching onto popular issues, and AI is the one place he has done the opposite, planting himself on the unpopular side of the fight over AI and data centers. That is a trade, not a miscalculation: the administration absorbs the political cost of backing an unpopular buildout, and in exchange the industry gets the runway to keep expanding data centers and deploying products the public has not consented to — whether that public is a Knoxville resident objecting to a camera on their street or the broader electorate that, per that same reporting, does not want any of this either.

Put the four pieces next to each other and the pattern is not subtle. A company accused of using AI to manufacture consent at the neighborhood level for the AI surveillance product it is selling. A lab admitting, in public, that its models keep behaving in ways it did not intend — and choosing to log the problem rather than halt the product. A king lending his stage to industry executives so they can be seen taking safety seriously, with nothing binding attached. And an administration willing to spend real political capital defending the same technology against a public that does not want it. None of these actors need to win the argument. They only need enough manufactured crowds, enough tracked-but-unaddressed incidents, enough royal photo ops, and enough political cover from Washington to keep building regardless of what residents, or voters, actually say.

The cost of that arrangement does not fall on Flock, or OpenAI, or the administration bearing it. It falls on the people in Knoxville who now have to fight not just a camera network but an astroturf campaign built to make their opposition look like the minority view — and on everyone living downstream of a technology whose own makers have publicly logged behavior they did not intend, and moved on.