There’s a courtroom in Oakland, California, where the most valuable idea in modern technology is currently on trial. Not a patent, not an algorithm, not even a specific product. The idea itself: that the best way to build a business is to build something people cannot put down.

The trial started on August 18, and it’s expected to run about six weeks in front of U.S. District Judge Yvonne Gonzalez Rogers. Four states are leading the charge – California, Colorado, Kentucky, and New Jersey – with 29 states in total accusing Meta of illegally hoovering up data on kids under 13. California deputy attorney general Megan O’Neill boiled the whole case down to one sentence in her opening statement: Meta’s business model was to “hook the users, hold them for as long as they can, harvest their data, and then hide the truth from the public.” Then she added the part that matters most. “It worked especially well for kids.“

Meta’s lawyer, Paul Schmidt, stood up and said the documents were cherry-picked out of longer conversations, that research shows no clear link between teen social media use and poor well-being, and that Meta’s products are not addictive. Both things can be argued for six weeks. But the deeper question hanging over the room is older and weirder than any lawsuit, and it goes something like this: what happens when a company gets very, very good at giving three and a half billion people a small chemical reward, over and over, all day long?

The quote that never went away

If you want to understand how we got here, start with Sean Parker in 2017. Facebook‘s founding president, the guy who helped make the thing, sat down with Axios and explained the design philosophy out loud. The goal, he said, was answering one question: “How do we consume as much of your time and conscious attention as possible?” And the answer was that “we need to sort of give you a little dopamine hit every once in a while, because someone liked or commented on a photo or a post.” He called it a social validation feedback loop. Then came the line that has been quoted in roughly every lawsuit since: “We’re exploiting a vulnerability in human psychology. The inventors, creators, it’s me, it’s Mark, it’s all of these people, understood this consciously. And we did it anyway.“

That’s the thing about the dopamine framing. It didn’t come from critics. It came from inside the building, offered up almost casually, as a description of good engineering.

The neuroscience behind it is real but often flattened in the retelling. Dopamine isn’t a pleasure chemical so much as an anticipation chemical. It spikes on the possibility of a reward, not the reward itself, which is why an unpredictable payoff is more compelling than a guaranteed one. It’s the mechanic that makes slot machines work, and plaintiffs’ lawyers have leaned on that comparison hard, arguing that infinite scroll and notification pings are variable reward schedules with a login screen. Meta’s response is that a feed is not a casino, and that likes, filters, and scrolling are, in the company’s word, benign.

What the internal documents say

The problem for Meta is not the theory. It’s the paper trail.

O’Neill told the jury that Meta employees described Instagram as a “drug” and themselves as “pushers,” and that an internal email to Instagram chief Adam Mosseri identified “teen time spent” as the goal. Other documents surfaced in court included one noting that “the young ones are the best ones” and another stating flatly: “Teens are hooked despite how it makes them feel. Instagram is addictive.“

Then there’s Project Mercury, which may be the single most damaging story in this whole saga. In 2020, Meta partnered with survey firm Nielsen to study what happened when people deactivated Facebook for a week. The results, according to court filings, were not what the company wanted: participants reported less depression, less anxiety, less loneliness, and less social comparison. Rather than publish or dig deeper, the filings allege, Meta shut the project down and blamed the unfavorable findings on the media narrative swirling around the company at the time. One researcher reportedly wrote that “the Nielsen study does show causal impact on social comparison,” followed by a frowning face. Another employee compared the decision to withhold the results to the tobacco industry sitting on what it knew about cigarettes. Meta spokesperson Andy Stone said the study was killed because of methodological flaws, and that the full record will show a company that spent more than a decade listening to parents and making real changes.

The states’ first witness has been Arturo Béjar, a former Meta engineering director who worked on safety from 2009 to 2015 and came back as a contractor on Instagram’s well-being team from 2019 to 2021, partly because of what he watched happen to his own teenage daughter on the app. Béjar told the jury that video autoplay, like counters, and infinite scroll are “inherently unsafe for teenagers,” and that Meta’s safety tools were “designed to fail” because they were optional rather than default. His analogy for Instagram’s quiet mode was sharp enough to stick: it’s like having “to turn on the air bag every time you get into the car.” He described the company’s approach to under-13 users as “don’t ask, don’t tell,” and said Reels shipped into the world with safety as an afterthought. On Zuckerberg specifically, he was blunt: “If Mark makes something a priority, mountains move in months.“

Meta’s cross-examination did land some hits. Béjar conceded that the company employs hundreds of safety people, many of them talented, and that he never actually solved the problems he was hired to solve. That’s the shape of the whole trial in miniature: not a story about villains, but about a machine whose incentives kept winning.

The uncomfortable part: the science is genuinely contested

Here’s where the honest version of this story gets harder. The “social media rewires kids’ brains” claim, as popularly told, is not settled science. The National Academies of Sciences, Engineering, and Medicine reviewed the literature and concluded it “did not support the conclusion that social media causes changes in adolescent health at the population level.” Candice Odgers, a psychologist at UC Irvine, argued in Nature that popular accounts of a phone-driven mental health epidemic mistake correlation for causation. Oxford’s Andrew Przybylski has spent years pointing out that Silicon Valley insiders sometimes accept the addiction framing more readily than the evidence justifies.

Meta leans on all of this, and on federal survey data showing major depressive episodes among 12 to 17 year olds fell from 21% in 2021 to 15% in 2024, arguing that teen mental health improved while social media use held steady or grew.

But notice what that defense does and doesn’t cover. You can accept that the population-level causal claim is unproven and still find it damning that a company allegedly ran the experiment, got an answer it didn’t like, and buried it. Juries have been noticing. In March, a Los Angeles jury found Meta and Google negligent and awarded $6 million to a 20 year old known in court as Kaley, with $4.2 million assigned to Meta. Zuckerberg testified in that case about lifting a temporary ban on beauty filters despite internal warnings about teen girls, saying, “I felt like the evidence wasn’t clear enough to support limiting people’s expression.” New Mexico has hit Meta with $942 million in penalties across two phases, plus mandated safety changes. And Breathitt County, a small rural Kentucky school district that was picked as the first of more than 1,200 district cases to reach trial, settled for a reported $27 million across the companies rather than going to a jury.

Meanwhile, the machine got better

The strangest thing about this moment is that while Meta argues in court that its products aren’t engineered for compulsion, it’s telling investors that its products are getting dramatically better at holding attention.

On the Q2 2026 earnings call, Zuckerberg said global time spent on Instagram grew by double digits year over year, “largely driven by improvements to our feed and Reels recommendations.” Meta now feeds every public Reel and post into a large language model to analyze topic and tone, which Zuckerberg called “a key building block toward greater personalization.” Revenue came in at $60.8 billion, up 28%, with 3.6 billion people using Meta’s apps daily. Buried in the same quarter: $2.4 billion in legal charges, and CFO Susan Li warning investors about continued scrutiny on youth issues.

That is the core tension, stated in the company’s own numbers. The engagement engine is the product, the profit center, and the legal liability, all at once. And it’s now generating its own fuel. Vibes, the AI video feed Meta launched in late 2025, was mocked almost immediately as an “infinite slop machine” of machine-made clips. Whatever you think of the quality, the direction is clear: a feed no longer limited by how much content humans can make.

Regulators are swinging, mostly with mixed results. Australia’s under-16 ban deactivated roughly 4.7 million teen accounts in its first month, but three months in, more than 80% of under-16s said they were still using social media anyway.

Which is why the Oakland remedies are worth watching more closely than the dollar figure. The states want the court to consider eliminating likes, killing infinite scroll, imposing time limits for younger users, and actually enforcing the under-13 rule. The money is enormous on paper: Meta says the exposure could reach $1.4 trillion, the attorneys general have floated something closer to $200 billion, roughly three years of after-tax profit. Meta’s stock fell 4.4% on the trial’s opening day. Kentucky attorney general Russell Coleman has been explicit about the playbook: “We did it with the tobacco settlement in the 1990s. We did it with the companies behind the opioid crisis. We’ll do it again with Meta.“

Money Meta can pay. What it cannot easily do is redesign the feedback loop that made it one of the most profitable companies in history. That’s the dopamine problem, and no verdict fixes it. A jury will hand Judge Gonzalez Rogers an advisory verdict in a few weeks, and she’ll decide liability and penalties herself. Then there will be appeals, possibly for years, possibly all the way up. And through every one of them, the scroll will keep going, faster and smarter than the year before, because that’s what it was built to do.