Meta just told investors it will pay roughly $9.89 billion to settle claims that it addicted teenagers, and the stock-market response to that sentence is a shrug. Revenue rose 26% to $64.6 billion. Earnings per share, charge included, came in at $3.83. Executives talked about the next product.
That is the real story. A penalty large enough to alarm any normal business was absorbed in a single quarter by a company still growing at a rate most CEOs would kill for. Call it what it is: a licensing fee. If harming children's attention and mental health is profitable enough that a near-$10 billion bill leaves you with a record-setting quarter, the fine did not punish the conduct. It priced it.
Notice also what the earnings summary does not say. There is no word about whether the engagement machinery that state attorneys general targeted, the infinite scroll, the notifications, the algorithms tuned to keep a fifteen-year-old on the app at midnight, will change. Money moved. Whether the product changed is a separate question, and Meta's presentation tells you where its attention lies.
Its attention is on Muse. More than 10 million people in the U.S. and Canada built personal AI agents in six weeks, Zuckerberg said, and Meta intends to make money through e-commerce fees, subscriptions for heavy users and enterprise sales. Consider what that means. An agent that knows your habits, your worries and your budget, built by a company with a documented history of optimizing for compulsion, will sit between you and what you buy, and Meta will take a cut of the transaction. Who is that assistant working for? A real fiduciary has duties. A salesman with a friendly voice has a commission. Meta is building the second and branding it as the first.
The pattern is familiar. Social media sold us connection and delivered an attention auction. Now the pitch is a helpful companion, and the business model is the same: capture intimacy, convert it to revenue, apologize later if the attorneys general come. This time the intimacy runs deeper, because people confide in these agents.
Then there is the plan for a non-voting Class C stock in January 2027, to fund acquisitions, pay employees and raise capital. Meta has said it will spend $175 billion to $179 billion next year, and it expects $15 billion in losses at Reality Labs alone. Those are sovereign-scale sums, committed on the judgment of one man who already controls the company's votes. A new class of non-voting shares lets him raise more money and issue more currency without surrendering any power. Shareholders get the privilege of financing the vision, not steering it. The public gets even less. Corporate democracy at Meta is already a formality, and this makes it a decorative one.
My own side should not escape. The settlement exists largely because state attorneys general did what Congress would not. Years of hearings, bipartisan indignation and televised scoldings of tech executives produced no durable federal rules for how platforms design products for minors. Democrats have been fluent in outrage and thin on legislation. Litigation became the substitute for lawmaking, and litigation produces cash and silence rather than rules.
The lesson is not that Meta should be congratulated for paying. It is that fines on a business this large are weather, not law. What works is binding design requirements for minors, real transparency about how recommendation systems operate, and a hard line separating an AI agent that advises you from one that sells to you. The next product is already in 10 million hands, and the rules still sit where they were before the last scandal.
How it may affect me
If you or your kids use Meta's apps, treat Muse like a salesperson, not a neutral assistant: Meta says it will earn e-commerce fees. Compare any product it recommends elsewhere, check which parental and teen-account settings are available, and be cautious about sharing financial or health details with an AI agent.
