The families and school districts suing Meta and Google say the platforms were built to hook children and that the companies knew what would follow. Meta's and Google's own insurers now say the same thing. They have to. A general liability policy pays for accidents, and the only way out of these cases is to persuade a court that nothing about them was accidental.
Two coverage suits over the same litigation landed in front of the same Delaware judge this year. In the first, Hartford and Chubb filed first, kept their forum, and won a ruling that they owe Meta no defense. In the second, Google filed first and kept its forum, so AIG will have to make the same argument in Santa Clara County. Both rulings came from Judge Sheldon K. Rennie of the Delaware Superior Court's Complex Commercial Litigation Division.
The forum rulings decide where the coverage fights happen. The Meta ruling decides what the fights are about, and it hands the plaintiffs an odd kind of win. A court has read their complaints and agreed that the conduct they describe was deliberate and its results foreseeable. The same reading takes the insurance money out of the case.
California's accident test
A general liability policy covers bodily injury caused by an "occurrence," and most policies define an occurrence as an accident. California courts use a two-step test to decide whether a lawsuit alleges one. Every coverage ruling in this piece turns on that test.
The first step asks whether the complaint alleges anything other than deliberate conduct. Conduct is deliberate if the defendant meant to do what it did, even if it never meant to hurt anyone. The second step applies only when the conduct was deliberate. It asks whether some independent and unforeseen event intervened to cause the harm. If the harm was a foreseeable result of the deliberate act, there was no accident and no coverage.
What the Delaware court held for Meta
Hartford Casualty and Sentinel sued Instagram and Meta in November 2024 in Hartford Casualty Insurance Co. v. Instagram, LLC, C.A. No. N24C-11-010-SKR CCLD. Chubb's Federal, Westchester and ACE units joined, along with more than a dozen other carriers as third-party defendants. The policies date to 2004, when Hartford wrote primary coverage for "The Face Book, Inc."
The insurers moved for partial summary judgment on the duty to defend. Meta moved to dismiss or stay in favor of a later-filed California action. On February 27, 2026, the court denied Meta's motion and granted the insurers'. The parties agreed California law governed the coverage question, and the court applied it.
Every policy at issue required an occurrence, and the parties agreed the term means an accident. The insurers' brief read like a plaintiffs' brief. Meta chose its algorithmic ranking, notification engineering and engagement optimization on purpose, they argued, and compulsive use by minors was the predictable result. Meta's subjective intent to harm anyone was beside the point. The insurers needed only to show that Meta intended the conduct that produced the harm.
The court resolved both steps of the accident test against Meta. The underlying complaints, read as a whole, attack Meta's intentional design of features built to maximize engagement. Addiction and the mental health injuries that followed were, in the court's reading, the foreseeable results of that design.
Negligence labels did not save the claims
Meta's best argument was that the complaints plead negligence, and negligence sounds like accident. The posture is worth pausing on. In Los Angeles and Oakland, Meta argues that its design harmed no one. In Wilmington, to keep its insurance, Meta argued that the same design is best read as a careless mistake.
The court looked past the label to the facts alleged. A negligent design count that describes deliberate design decisions describes deliberate conduct. Allegations that Meta "should have known" of the risks describe the foreseeable consequences of those decisions.
The failure-to-warn counts fared no better. The court treated them as misrepresentation claims, which California law does not treat as accidents. The opinion noted that failure-to-warn facts might support a separate negligence theory, but Meta did not press the argument. Third-party content did not break the chain either. As the complaints describe the platforms, that content is what the design was built to deliver.
A liability policy gives the insurer two separate jobs. The duty to defend requires it to pay the policyholder's lawyers while the lawsuits are pending. The duty to indemnify requires it to pay the judgments and settlements that come out of them. The insurers asked the court to rule only on the first duty, so the court decided only that Meta's insurers do not have to pay for Meta's lawyers.
The court left open whether the insurers must pay any judgment or settlement. That opening gives Meta little room. Under California law, an insurer that owes no defense usually owes no indemnity either. The court also rejected Meta's argument that the plaintiffs might later amend their complaints in a way that triggers coverage. An amended complaint would have to plead new facts to change the result.
Google's insurers will make the same argument in California
Google's dispute is with American Home Assurance Co. and National Union Fire Insurance Co. of Pittsburgh, both AIG units. Alphabet began tendering the social media claims to AIG in July 2024. AIG reserved its rights that November. According to Alphabet, AIG said in February 2025 that a coverage decision would come within three weeks. A year later, Alphabet still had no answer.
| Date | Event |
|---|---|
| July 2024 | Alphabet begins tendering claims to AIG |
| November 2024 | AIG reserves its rights |
| February 2025 | AIG says a coverage decision will come in three weeks, per Alphabet |
| February 2026 | Alphabet, Google and YouTube sue AIG units in Santa Clara County |
| February 27, 2026 | AIG units sue in Delaware; the Meta coverage ruling issues the same day |
| June 2026 | AIG issues a coverage letter |
| September 21, 2026 | Delaware court stays the AIG case in favor of California |
The Delaware suit, American Home Assurance Co. v. Google, LLC, No. N26C-02-652, seeks a declaration that AIG owes no defense or indemnity. Judge Rennie found the California case was filed first, involves substantially the same parties, and can resolve the contract questions in full. He rejected AIG's argument that the California court was moving too slowly. He also found no gamesmanship in Google's filing.
Scott Seaman of Hinshaw & Culbertson practices on the insurer side. He called Google's 16-day head start "game, set, and forum match." He told Law360 that with the Meta appeal pending, the outcome of both coverage actions "still hangs in the balance." In a follow-up on his firm's site, he wrote that the insurers have strong arguments on the merits and may well prevail in California.
Read from the plaintiffs' side, that is a coverage lawyer predicting that a California court will accept their account of how the platforms were built.
What changes in California
Less than the forum fight suggests. California law governed the Meta ruling, and California law will govern the Google case. A Santa Clara judge will apply the same two-step accident test to complaints that make the same design allegations against YouTube that the Delaware court read against Instagram.
What changes is the decision-maker and the appellate path. A Delaware trial judge's reading of California law binds no California court. Any appeal in the Google case goes to California's Sixth District Court of Appeal, which can read the accident cases its own way. If the Delaware Supreme Court reverses in Meta, Google gains a persuasive opinion. If it affirms, AIG gains one.
Google also has a fact Meta lacked when its case was briefed. The first bellwether in the California coordinated proceeding, JCCP 5255, went to verdict on March 25, 2026. A Los Angeles jury found Meta and Google negligent in the design of their platforms, in a case brought by a plaintiff identified as K.G.M. Google will argue that a jury has now found negligence, and negligence is the language of accident.
AIG's first answer comes from the Meta ruling. The Delaware court held that a negligence label does not turn deliberate design into an accident. A negligence verdict on a design theory has the same problem.
What the punitive award tells the coverage court
The K.G.M. jury did more than find negligence. It awarded $3 million in compensatory damages and another $3 million in punitive damages. The $6 million total was split 70% to Meta and 30% to Google, so Google's share came to $1.8 million. For the coverage fight, the punitive award may matter more than the negligence finding.
California allows punitive damages only on clear and convincing proof of malice, oppression or fraud, under Civil Code section 3294. The statute defines malice to include despicable conduct carried out in willful and conscious disregard of the safety of others. A jury that awards punitive damages has found conduct of that kind, under a stricter standard of proof than the rest of the case required.
That finding is hard to square with an accident. Conduct carried out in conscious disregard of a known risk is deliberate conduct with a foreseeable result. Under the accident test, deliberate conduct with a foreseeable result is no occurrence. The K.G.M. jury called the design negligent, and in the same verdict it found the conduct behind the design worse than negligent.
The punitive dollars carry a second problem for Google. California public policy bars insurance for punitive damages, under Peterson v. Superior Court (1982). Unless the award is overturned on appeal, the punitive share stays with Google whatever the coverage court decides. Insurance Code section 533 separately bars indemnity for a loss the insured caused willfully, and a malice finding is evidence AIG will use under that statute too.
The insurers' argument is the plaintiffs' argument
Section 230 bars claims that treat a platform as the publisher of someone else's content. Plaintiffs in the federal MDL and the California JCCP built their cases around platform design to get past it. Infinite scroll, autoplay, engagement algorithms and notification timing are the platform's own conduct. Courts in both proceedings let design-based claims go forward while trimming claims aimed at what users posted.
The allegations that carry a design claim past Section 230 are the allegations the insurers used to escape coverage. A complaint that says the defendant engineered its product to hook children pleads deliberate conduct with foreseeable harm. It also pleads the malice that supports punitive damages. Under California's accident test, that same complaint pleads no occurrence. The insurers did not have to write a new theory of the case. They adopted the plaintiffs' theory and filed it in Delaware.
A coverage ruling is not a merits finding. The Delaware court read the complaints as pleaded and decided what kind of conduct they describe. It did not decide whether Meta did any of it, and nothing in the opinion binds the juries in Los Angeles or Oakland. What the plaintiffs gain is narrower and still useful. The carriers that wrote Meta's policies, with their own money at stake, looked at the design allegations and told a court they describe deliberate conduct. The court agreed.
Building the cases on design worked. Design claims survived Section 230, and a design claim produced the K.G.M. verdict. The cost of the design theory comes due when the families and school districts suing Meta and Google try to collect.
In most product cases, the defendant's general liability insurers pay for the defense. They also fund much of any settlement or judgment. Large companies stack several of these policies, from a primary policy up through umbrella and excess layers.
A complaint alleging that the defendant built its product to hook children describes deliberate conduct. Under the Delaware court's reasoning, every insurer in that stack can decline the defense, and an insurer that owes no defense usually owes nothing toward the payout. Whatever the plaintiffs recover then comes from the defendant's own money.
Who the ruling hurts
For the plaintiffs suing Meta and Google, the answer is close to nobody. In August, Meta agreed to pay up to $18 billion to settle claims brought by nearly every state. About $12.7 billion is payable over ten years, and another $5.3 billion depends on other platforms reaching similar deals. YouTube settled with the plaintiff set for the second JCCP trial, just before that trial in June. These defendants pay settlements and defense costs from the balance sheet, with or without a carrier.
The ruling matters more for the defendants that come next. The video game addiction cases borrow the social media design theory, and so do the AI chatbot wrongful death cases. Many of those defendants are smaller companies. If their general liability carriers deny a defense on day one, defense costs draw down the same balance sheet that would fund a settlement.
A plaintiff firm valuing inventory against a smaller platform should ask early about coverage. The relevant documents are the tender, any reservation of rights letter, and any declaratory action the carrier has filed. Other lines of insurance, such as technology errors and omissions policies, use different insuring language. The Meta ruling addressed general liability policies only.
What to watch
The Delaware Supreme Court's decision in Meta comes first. An affirmance would give every general liability carrier in the platform cases a high-court opinion reading California law their way. A reversal would revive Meta's defense claim and undercut AIG's position in Santa Clara.
The Santa Clara action is the second test. It is the first chance for a California judge to apply California's accident test to the social media complaints. The duty to indemnify is the third. No court has ruled on it yet, including in Meta. The K.G.M. verdict and any settlement payments give insurers and policyholders a fixed set of facts to argue over.
The plaintiffs are not parties to any of it, and they should read all of it. Every brief the carriers file is a statement, by a party with money on the line, that the platforms built their products on purpose and knew what would follow. The plaintiffs have been saying that since 2022. They now have company.