Social media giant Meta has agreed to pay up to $18 billion to 48 US states, the District of Columbia, and three territories to settle claims that its Facebook and Instagram platforms harmed children. The deal, announced Wednesday and subject to approval by a California federal court, marks Meta’s largest-ever payment over child safety litigation and was reached after only five days of trial in an Oakland courtroom. Under the settlement, Meta will implement sweeping platform changes: default two-hour daily time limits for teen users (adjustable only by a parent), night-mode notifications blocked between midnight and 6am, hidden likes on teen accounts, school-hours notification muting, and removal of extreme make-up filters. The daily limit will drop to one hour if rival platforms including TikTok, Snap, and YouTube agree to equivalent restrictions. California Attorney General Rob Bonta called it “a major moment to clean up an industry that has been hurting our kids.” Meta denied wrongdoing but agreed the $18bn would be paid in annual instalments over ten years. The settlement does not include New Mexico, where a federal judge last month found Meta to be a “public nuisance” and ordered it to pay nearly $1 billion in fines.
The received wisdom
The mainstream interpretation of this settlement is essentially correct as far as it goes: this is a significant win for public accountability, and the internal evidence presented during the brief trial was damning. Internal Meta research stated flatly that “teens have an addict’s narrative about use.” A former researcher, George Volichenko, testified that a manager had told him not to worry about low adoption of safety features “because the team exists partially to protect the company against the upcoming lawsuits.” The company apparently knew what it was doing to young users and made the calculated judgment that monetisation was worth the exposure. The states’ lawyers argued Meta knew millions of 11- and 12-year-olds were using Instagram and Facebook and “did little to keep them off.” On those facts, the progressive position — that this was a predatory industry extracting profit from children’s psychological vulnerabilities — deserves to be taken seriously. Public-health analogies to tobacco and opioids are not rhetorical excess; they are grounded in the internal communications that surfaced during discovery.
A different read
The question is whether this settlement, for all its headline grandeur, actually changes the incentive structure that produced the problem — or whether it simply prices the harm into the business model.
$18 billion is a large number. Spread over ten years, it is $1.8 billion a year. In the quarter just reported, Meta’s revenues were running at a rate that would imply roughly $180 billion annually. The settlement, in other words, amounts to approximately one percent of annual revenue. The tobacco master settlement agreement, signed in 1998, was also celebrated as a landmark. It too was spread over decades. What it produced was not the structural reform of the tobacco industry but a stable funding stream for state governments that became, perversely, dependent on tobacco sales continuing. State attorneys general who negotiated the deal later found themselves in the odd position of having fiscal interests aligned with the industry they had prosecuted.
The pattern here is not identical, but the structural risk is the same. If Meta can budget for child-safety settlements as a predictable line item — a regulatory cost, like spectrum fees or environmental compliance — then the incentive to genuinely redesign its products is limited. The promised platform changes are real and not trivial: default time limits and night-mode blocking are more substantive than the industry’s previous voluntary commitments. But they are also controllable, adjustable, and — critically — only as robust as Meta’s enforcement of them.
The BBC’s technology editor noted that the settlement may hasten a “reckoning” for social media on child safety more broadly, and California’s Bonta explicitly invited TikTok and YouTube to adopt the same framework. This is the more interesting leverage point: if the settlement creates a sector-wide standard, enforced through subsequent litigation against non-compliant platforms, the impact compounds. But “urging” TikTok to comply is not the same as requiring it, and TikTok’s relationship with the US regulatory apparatus is complex enough that such an invitation may go unheeded.
What the settlement does not address is the deeper architecture of engagement optimisation: algorithmic systems that are not specifically designed to harm teenagers but are designed to maximise time-on-platform, and which happen to work with particular efficiency on adolescent neurology. Two-hour daily limits are a constraint imposed on top of an unchanged underlying system. A more structurally reformist settlement would have required changes to the recommendation engine itself — not just the duration of exposure, but the nature of the content surfaced to vulnerable users. That was not on the table, presumably because it touches Meta’s core commercial product rather than its peripheral safety features.
There is a conservative case for this kind of robust product-level regulation, and it is distinct from the progressive public-health framing. It rests not on the precautionary principle but on a straightforward application of liability: if a product is knowingly designed in a way that causes harm to minors, the manufacturer is liable, and the appropriate remedy should reflect the cost of the harm caused, not merely a negotiated fraction of the company’s revenue. The tobacco analogy holds at this level: what ultimately changed tobacco was not the 1998 settlement but the discovery process that produced it — the internal documents that made denial impossible — and the litigation that continued after it. The Meta settlement opens that discovery process to public view. Whether legislators will use it to push for structural reform, or whether the $18bn will simply close the chapter, depends on political will that the settlement itself cannot supply.
What to watch
- Whether TikTok, Snap, and YouTube adopt equivalent teen-safety frameworks — and whether Bonta’s office follows through on its signal of “more focused” attention on non-compliant platforms.
- The New Mexico litigation: unlike the multi-state settlement, the “public nuisance” ruling is a judicial finding, not a negotiated deal, and it could be the vehicle for more structural product-design orders.
- Whether the EU’s Digital Services Act enforcement, moving on a parallel track, produces stricter algorithmic transparency requirements that would effectively set a global standard Meta could not avoid.
- The political metabolism of the internal documents now in public record: Congressional committees have a trove of evidence that the last round of Big Tech hearings lacked.
— J