Meta’s design features face court as states seek product changes

Four state attorneys general opened a federal trial against Meta Platforms in Oakland on Tuesday, accusing the company of designing Facebook and Instagram to keep children and teens scrolling while publicly downplaying the risks. California Deputy Attorney General Megan O’Neill told the advisory jury that Meta’s model boiled down to four words: hook users, hold them, harvest their data, and hide the truth.

“It was especially bad for kids,” O’Neill said. The case, litigated by California, Colorado, Kentucky and New Jersey on behalf of a larger 29-state coalition, centers on product design rather than user posts, a choice that lets the states sidestep long-standing legal shields.

Opening statements set the clash over profits and safety

O’Neill pointed to a 2016 internal email stating Instagram’s “overall company goal” was “teen time spent.” She cited a company study called “Long Term Retention: The Young Ones Are The Best Ones” that examined tweens roughly ages 10 to 12 and concluded earlier starters stayed longer and generated more revenue.

Meta said it prioritized safety over profits, but it hid the reality that when it came time to make a decision, time and again profits won.

That line came from O’Neill’s opening. Meta lawyer Paul Schmidt told the eight-member advisory jury the states were cherry-picking documents out of context. He said Meta has long discussed people struggling with social media and has taken meaningful steps to address it.

“Let me be clear: Meta has talked over the years about people struggling with social media,” Schmidt said. He acknowledged some teens struggle with time management and negative posts but insisted the company acts on those issues. Much of the suit, he argued, amounts to disagreement over how Meta improves and how it talks about improvement.

Meta said in a statement to NPR that the claims are unsubstantiated and that it stands by strong teen protections, including enhanced privacy settings and a one-hour Instagram timer. In a separate statement quoted by CNBC, the company called the AGs’ claims limited and their financial demands vastly disproportionate, adding that the states offer no proof anyone was misled and try to penalize Meta for industry-wide age-verification challenges.

The design theory that dodges Section 230

Social platforms have long relied on Section 230 of the Communications Decency Act and First Amendment arguments to avoid liability for user content. This trial takes a different route. The states say Facebook and Instagram were built with features that encourage compulsive use by minors, boosting ad revenue while harming mental health.

Features at issue include:

• Infinite scroll that removes natural stopping points
• Autoplay and ephemeral content that keep feeds moving
• Beauty filters and the like button that shape social comparison
• Engagement-optimized algorithms that surface more of what holds attention

The states also allege Meta violated the Children’s Online Privacy Protection Act rules by collecting personal data from users under 13 without verified parental consent. If they prove COPPA violations, they want Meta to delete under-13 personal data and the algorithms and models trained on it. For state consumer-protection claims they seek permanent nationwide injunctive relief removing certain addictive design features.

California Attorney General Rob Bonta Bonta laid out the states’ case the day before openings, saying Meta designed the products to keep kids on longer to the point of physical and mental harm. Colorado AG Phil Weiser said Meta knew the platforms could harm young people yet kept practices that sacrificed sleep and school focus because more time meant more money. Kentucky AG Russell Coleman compared the effort to the 1990s tobacco settlement and the opioid cases. New Jersey AG Jennifer Davenport called it endangering an entire generation with features Meta knew would have horrific effects.

Earlier verdicts already cost Meta nearly a billion

Meta has lost two related state-court fights in 2026. Those results supply the blueprint the Oakland trial is testing at larger scale.

New Mexico AG Raúl Torrez told CNBC the New Mexico result was substantial for a state of two million people and that mapping the same logic onto California or other large states could be market-shifting. He called potential California consequences astronomical. Meta has said it disagrees with the New Mexico rulings and will appeal.

Bonta told NPR before openings that if the trend continues Meta will lose again, pay a lot again, and have to make changes. He said the company can still be wildly successful without features that create mental-health harms. “We want Meta, in short, to stop hurting our kids, stop knowingly hurting our kids,” he said.

Whistleblower Arturo Béjar takes the stand first

Former Meta engineer Arturo Béjar, who worked in product safety across two stints totaling eight years, was the states’ first witness after Meta tried unsuccessfully to bar him. He testified that internal surveys of more than 200,000 users showed younger people reported higher rates of encounters with bullying, self-harm and violence for almost every issue.

Instead of publicizing those harm rates, he said, Meta released different metrics based on content-policy violations that did not equal harm. “I think that these numbers create a false impression of safety,” Béjar said. During his 2019-2021 return the company used the label “problematic use” rather than addiction. That label, he testified, undercounted what academic literature considers addiction.

Crowd discussion of the testimony noted Béjar’s personal stake: he returned after his 14-year-old daughter received sexual requests and unsolicited explicit images she could not easily report. He described safety as an afterthought at the time and said he raised concerns with senior leaders, including interactions with Mark Zuckerberg estimated at least 100 times. These details from live coverage and posts sharpened the picture of internal knowledge versus external messaging. Béjar’s testimony was set to continue Wednesday.

The $1.4 trillion figure and what the states actually want

An initial calculation submitted by the states and later challenged by Meta put possible maximum penalties near $1.4 trillion, roughly Meta’s Nasdaq equity value at the time. Bonta said the states are not seeking a specific number and accused Meta of highlighting the ceiling to make the case look unreasonable. He noted Meta generated about $200 billion in revenue last year and said that amount, or more or less, might be appropriate.

Meta argued in pretrial filings that the math multiplies maximum fines across every teen user without proof of separate affirmative acts or actual exposure, double-counts, and would be disproportionate under due-process precedent. The tobacco Master Settlement Agreement totaled $206 billion. States told the court $200 billion is a more likely figure. Judge Yvonne Gonzalez Rogers, who also handled Elon Musk’s suit against OpenAI, is presiding. The advisory jury will hear liability; the judge is expected to handle monetary relief.

Beyond dollars, the states want permanent injunctive relief on a nationwide basis for COPPA violations and removal of specific design features for consumer-protection violations. In New Mexico the judge declined some design mandates such as ending infinite scroll, citing potential Section 230 and First Amendment conflicts and fairness to rivals that keep the tools. That leaves open whether the federal court will go further.

Ad cash, teen time and the AI spending pressure

Meta reported in late July that Family daily active people averaged 3.60 billion daily active people and $60.8 billion revenue in the second quarter, up 28 percent year over year. Ad impressions rose 14 percent and average price per ad rose 12 percent. The company also booked $2.40 billion in charges related to legal proceedings in the quarter and flagged that youth-related trials scheduled this year may ultimately result in a material loss.

Capital expenditures are guided at $130-145 billion for 2026 as Meta builds AI capacity. That spending is funded largely by the advertising business that thrives on time spent. Analysts have focused more on the capex bill than on litigation risk, but Torrez argued Wall Street is underestimating a California judgment large enough to change Meta’s ability to finance its future. Any forced removal of engagement features or deletion of under-13 trained models would strike at the same metrics the 2016 email and retention study once celebrated.

The pressure lands while Meta is already navigating Meta’s massive AI infrastructure spending bet and drawing comparisons to echoes of Meta’s earlier metaverse spending cycle. A redesign order that reduces teen time-on-app would not merely add a fine; it would alter the raw material of the ad auction just as the company is writing the largest infrastructure checks in its history.

Six weeks of evidence and an open question on Zuckerberg

The trial is expected to run about six weeks. Attorneys general said Tuesday they had not yet decided whether to call Meta CEO Mark Zuckerberg as a witness. Legal observers expect Meta to appeal any adverse ruling, potentially to the Supreme Court. Parents and advocates gathered outside the courthouse to put a human face on the claims as openings began.

Bonta has framed the goal simply: stop the features that create known mental-health harms while still allowing a successful business. Meta frames the case as disagreement over improvement methods and language, plus unproven claims about benign features. The design documents, the survey numbers Béjar described, the prior state verdicts, and the scale of the AI capex now sit in the same courtroom. Whatever the advisory jury and judge decide, the second-order fight over who controls the scroll is already under way.