IN RE: SOCIAL MEDIA ADOLESCENT ADDICTION/PERSONAL INJURY PRODUCTS LIABILITY LITIGATION
Administrative Motion to File Under Seal RE MOTION IN… — Attachment 19
Issue: In *In re: Social Media Adolescent Addiction*, Meta Platforms argues, through Motion in Limine No. 3, that sworn trial testimony from a parallel New Mexico state-court proceeding — describing Meta's algorithm as automatically and predictably escalating users toward harmful content — should be excluded or restricted at the federal MDL trial. The non-obvious tension is that Meta itself attached the testimony as its own exhibit while simultaneously seeking to seal it, raising the question of whether a platform can control the evidentiary pathway through which damaging insider testimony enters a federal record without triggering the full weight of that testimony against it.
Meta filed an administrative motion to seal its Motion in Limine No. 3 and twenty-two accompanying exhibits in the Northern District of California MDL on June 9, 2026, during the pre-trial evidentiary motions phase. Among those exhibits is Exhibit Q — a transcript excerpt from the New Mexico *v.* Meta trial (Feb. 12, 2026) in which former Meta employee Brian Boland testified about the company's algorithmic recommendation system. Boland testified that Meta's machine-learning algorithm operates in a manner no one at the company fully understands, that a single user interaction with sensitive content triggers a documented, self-reinforcing escalation toward harmful material, and that an internal Meta document called "Carol's Journey to QAnon" — showing a new account algorithmically driven to fringe conspiracy content within weeks — was never voluntarily disclosed to regulators and reached Congress only through a whistleblower leak. Boland further confirmed that the same personalized feed driving this content escalation is Meta's primary advertising delivery mechanism. The underlying MIL seeks an evidentiary ruling governing whether this testimony and the related internal research documents may be used at the MDL trial.
Meta's decision to introduce this transcript as its own exhibit while seeking to seal it is the central strategic puzzle: the company appears to be attempting to manage how damaging insider testimony enters the federal record rather than allowing plaintiffs to introduce it on their own terms and framing. The sealing request is legally vulnerable under *Kamakana v. City & County of Honolulu* because the testimony comes from an open state-court trial and carries a strong presumption of public access, particularly where the underlying MIL bears meaningfully on the scope of trial. If the court denies sealing and resolves the MIL against Meta, Boland's acknowledged combination of documented internal knowledge of predictable harm pathways, the suppression of that knowledge, and the direct revenue linkage to the contested algorithmic design could substantially strengthen plaintiffs' corporate-knowledge, concealment, and punitive damages theories across the MDL.
MOTION in Limine NO. 3 TO EXCLUDE TESTIMONY FROM CERTAIN… — Attachment 3119
Issue: In *In re: Social Media Adolescent Addiction/Personal Injury Products Liability Litigation*, Meta Platforms argues that Section 230 of the Communications Decency Act — which the court applied at the motion-to-dismiss stage to immunize certain platform features — should now operate as a categorical bar on trial testimony about those same features, preventing eight former Meta employees from testifying about algorithmic recommendations, infinite scroll, autoplay, notifications, and related design choices. The argument is doctrinally unusual because it asks the court to treat a prior liability-immunity ruling not merely as a defense on the merits, but as a forward-operating evidentiary exclusion under FRE 401 and 403. Meta separately argues that the witnesses lack sufficient personal knowledge under FRE 602 and 701, and that their broad characterizations of corporate culture and executive priorities constitute impermissible lay opinion.
Meta filed Motion in Limine No. 3 on June 9, 2026, ahead of trial in this major multidistrict litigation over alleged harms to adolescents from social media platforms, with a hearing set before Judge Yvonne Gonzalez Rogers on June 26, 2026, and opposition due June 17, 2026. The motion targets eight named former Meta employees and seeks to exclude their testimony on two independent grounds. First, Meta argues that because the court's earlier motion-to-dismiss order (ECF 1214) held that certain platform features are immunized under Section 230, any testimony concerning those features is now irrelevant and unfairly prejudicial at trial. Second, Meta contends that deposition testimony from the witnesses reveals gaps in personal knowledge — for example, one witness allegedly admitted unfamiliarity with teen-safety initiatives, another never worked on Instagram — and that their broad opinions about Meta's institutional culture and leadership priorities go beyond what even qualified experts may offer. Meta also invokes a California state-court ruling from the parallel JCCP proceeding that reportedly limited one witness's testimony, urging the federal court to follow the same approach.
Meta is attempting something more aggressive than a typical motion in limine: it wants a prior legal ruling that it cannot be held liable for certain features to also mean that witnesses cannot testify about those features at trial — the difference between a court saying "you cannot win on that theory" and saying "the jury cannot hear about it at all." If the court accepts this theory, it could substantially narrow the plaintiffs' trial presentation on core design-defect claims that survived the motion-to-dismiss stage, because the platform features at issue are the same ones underlying the surviving theories. No circuit court appears to have squarely endorsed using a Section 230 immunity ruling as a prospective evidentiary exclusion, making Judge Gonzalez Rogers's ruling on this question potentially the first of its kind in a major platform-liability MDL — and a significant marker for how Section 230 doctrine functions once a case reaches trial.
MOTION in Limine NO. 2 TO PRECLUDE EVIDENCE… — Attachment 3117
Issue: In *In re: Social Media Adolescent Addiction/Personal Injury Products Liability Litigation*, Meta Platforms argues that evidence showing it *could have* adopted different age-verification systems, content-moderation practices, or platform-access controls is both legally irrelevant and independently barred by Section 230 and the First Amendment. The question is whether those three legal doctrines — operating together as a pre-trial evidentiary shield — can prevent a jury from ever hearing expert testimony about safer design alternatives, even in a government consumer-protection action premised on Meta's own statements and conduct rather than its handling of third-party content.
Meta filed this Motion in Limine No. 2 on June 9, 2026, in the California Attorney General subtrack of MDL No. 3047, with a hearing set before Judge Yvonne Gonzalez Rogers on June 26, 2026. The motion seeks a pre-trial order excluding all evidence, argument, and expert testimony suggesting Meta should have implemented different safety, age-verification, or moderation practices. Meta advances three independent grounds: that such evidence is irrelevant under FRE 401–403 because the AGs' claims turn on what Meta actually said and did; that Section 230 immunizes age-gating and moderation choices as protected "publisher decisions" under *Doe v. Grindr* (9th Cir. 2025); and that the First Amendment, as interpreted in *Moody v. NetChoice* (2024), forecloses evidence compelling Meta to justify its editorial and access-restriction choices. The motion specifically targets six named experts — Alter, Estes, Sheatsley, Gray, Iyer, and Bejar — seeking to exclude their opinions in their entirety on all three grounds simultaneously.
Meta is asking the court to block, before trial begins, the evidentiary foundation the California AG would use to show that Meta's conduct was unreasonable — the equivalent of barring a products-liability plaintiff from presenting evidence that a safer design existed. If granted, even partially, this ruling would signal that Section 230 and the First Amendment can together preempt the expert toolkit in state government enforcement actions, with immediate consequences for every AG suit currently pending against social media platforms. The motion also advances a largely untested procedural theory: that Section 230 can function as an *in limine* evidentiary bar, not merely a pleading-stage immunity — a doctrinal expansion that no circuit has squarely endorsed. Judge Gonzalez Rogers's ruling will be closely watched as a potential bellwether on whether platforms can use constitutional and statutory shields to hollow out liability cases without ever reaching a merits trial.
Proposed Order re [3117] MOTION in Limine NO. 2 TO… — Attachment 3125
Issue: In *In re: Social Media Adolescent Addiction/Personal Injury Products Liability Litigation*, Meta argues that five categories of anticipated State Attorney General evidence should be excluded at trial — covering omission theories, policy-critique expert opinions, former employee testimony, alleged misrepresentations, and age-verification practices — primarily on the grounds that they are irrelevant, unfairly prejudicial, or independently barred by Section 230 immunity and the First Amendment. The non-obvious difficulty is that several of Meta's proposed exclusions depend on legal positions — particularly that age-verification mechanisms are protected editorial decisions under Section 230, and that the AGs' prior disavowal of a duty-to-disclose theory operates as a blanket waiver of all omission evidence — that have no firm appellate support and conflict with established evidentiary principles and federal statutory frameworks.
Meta's counsel filed this document on June 9, 2026, as a party-drafted proposed order submitted for court adoption at the pre-trial stage of the California AG bellwether trial in MDL No. 3047 (N.D. Cal.). Although formatted in judicial language, it is an advocacy document — not a court ruling — presenting Meta's preferred resolution of its five pending Motions in Limine. MIL 1 seeks exclusion of all omission evidence based on the AGs' earlier litigation representations that they are not pursuing a standalone duty-to-disclose theory. MIL 2 targets six named expert witnesses whose design-alternative opinions Meta characterizes as irrelevant policy second-guessing independently barred by Section 230 and the First Amendment. MIL 3 seeks categorical exclusion of eight named former employees from testifying about algorithms, engagement features, content moderation, and corporate culture, invoking FRE 402, 403, 701, and 702. MILs 4 and 5 seek exclusion of evidence tied to 26 specific alleged misrepresentations and to age-verification and third-party filter practices, with Meta arguing both categories are immunized by Section 230 or forfeited by the AGs' litigation conduct.
This filing is worth watching because it maps the outer boundaries of how Meta intends to deploy Section 230, the First Amendment, and evidentiary forfeiture arguments simultaneously to foreclose the State AGs' trial presentation before it begins. The most consequential and legally unsettled move is Meta's effort to extend Section 230 immunity to age-verification mechanisms by characterizing them as publishing decisions — a position that directly conflicts with COPPA's knowledge-based framework and that no circuit court has endorsed. If a trial court adopted that framing, it would generate a significant appellate question about whether Section 230 can insulate platforms from evidentiary scrutiny of their age-detection capabilities in the face of a federal statutory scheme that specifically contemplates that inquiry. The proposed blanket waiver of omission evidence and the categorical exclusion of former employee percipient testimony present additional doctrinal pressure points, since both arguments run against well-established principles that such evidence routinely remains admissible to contextualize affirmative claims and satisfy the personal-knowledge requirements of FRE 701.
NOTICE by Meta Platforms, Inc. OF FILING OF FOLLOW-UP… — Attachment 12
Issue: In *In re: Social Media Adolescent Addiction/Personal Injury Products Liability Litigation*, Meta Platforms argues that the State Attorneys General cannot credibly claim surprise or prejudice from five of Meta's trial witnesses because the AGs' own October 2025 interrogatory responses demonstrate they had full notice of those witnesses' relevance and subject matter. The underlying legal question — whether Meta adequately disclosed these witnesses during discovery such that striking them is warranted — turns on the interplay between MDL case management orders, Rule 26 disclosure obligations, and the doctrine that a party cannot claim prejudice from information it already possessed.
This is a Notice filed by Meta on June 4, 2026, at the pre-trial stage of the Social Media MDL in the Northern District of California, in connection with Meta's opposition to the State AGs' motion to strike five of Meta's witnesses from its trial witness list (ECF 3056). The filing attaches twenty exhibits, anchored by Exhibit J — the AGs' own Third Supplemental Interrogatory Response from October 2025 — which Meta deploys to show that the AGs had detailed knowledge of the contested witnesses' anticipated subject matter, including platform design, age verification, COPPA compliance, and data practices regarding users under thirteen. Meta simultaneously requests that the AGs' own supplemental filing (ECF 3104) be stricken as a reply-style submission filed without leave of court in violation of local rules. The filing is styled as a neutral Notice but functions substantively as a follow-up opposition brief, with the voluminous exhibit record marshaled to demonstrate that the five witnesses' anticipated testimony tracks issues the AGs themselves placed at the center of the litigation.
This pre-trial skirmish over witness lists telegraphs the shape of the trial itself: the five contested witnesses are positioned to contest or contextualize the core claim that Meta designed its platforms with knowledge that they were harmful to children, which maps directly onto the negligent design and platform-own-conduct theories that remain unsettled across the broader § 230 landscape. Meta's strategy of deploying the AGs' own detailed interrogatory allegations to win a procedural point carries a notable risk — those same allegations describe Meta's purported knowing harm to minors in terms that may be difficult to cabin once introduced into the trial record for any purpose. For observers tracking how product-design defect claims against social media platforms will be tested against actual trial evidence rather than pleadings, this filing is a useful preview of the evidentiary terrain on which those doctrines will be litigated in what is among the highest-stakes platform liability proceedings currently in active pre-trial proceedings.
NOTICE by Meta Platforms, Inc. OF FILING OF FOLLOW-UP… — Attachment 5
Issue: In *In re Social Media Adolescent Addiction/Personal Injury Products Liability Litigation*, Meta Platforms argues that five witnesses it intends to call at trial were timely and adequately disclosed under the applicable pretrial schedule, and that the State Attorneys General's motion to strike those witnesses mischaracterizes the disclosure record. A secondary question runs in parallel: whether the AGs' own supplemental filing (ECF 3104), submitted without apparent leave of court, was itself procedurally improper and subject to being stricken. Both questions turn on how strictly the MDL court is enforcing its case management orders as trial approaches.
Meta filed this notice on June 4, 2026, in the Northern District of California MDL, in response to the State AGs' motion to strike five Meta witnesses from Meta's trial witness list (ECF 3056) and the AGs' subsequent supplemental filing on that motion (ECF 3104). Although styled as a "Notice of Filing Follow-Up Materials," the document functions substantively as a supplemental brief in support of Meta's earlier opposition (ECF 3060-1), accompanied by 22 exhibits and a declaration from Meta's counsel Ashley M. Simonsen authenticating those exhibits. The exhibits are offered to establish the communications, correspondence, and prior filings that Meta contends demonstrate both adequate witness disclosure and the AGs' procedural overreach. Embedded within the notice, Meta also affirmatively requests that the Court strike ECF 3104 on the ground that the AGs filed additional argument without leave, in violation of local rules and the Court's briefing schedule. Meta's position is that striking its witnesses on the eve of trial would impose disproportionate prejudice given the existing discovery record, while the AGs' unauthorized supplemental brief compounds that unfairness by injecting new arguments outside the agreed schedule.
This filing is a procedural skirmish, not a substantive legal development, but it carries practical significance for how the trial will unfold: whether five Meta witnesses survive or are stricken will directly shape the evidentiary landscape at trial. The mutual accusations of improper supplemental filing — each side effectively doing what it accuses the other of doing — illustrate how pre-trial witness disputes in complex MDLs can become procedural warfare that runs independent of the underlying merits. The Court's response will serve as a reliable signal of how tightly it intends to enforce case management discipline in the final stretch before trial, and practitioners monitoring this MDL should treat the ruling as an early indicator of the judge's trial-phase posture rather than as doctrinal output.
NOTICE by Meta Platforms, Inc. OF FILING OF FOLLOW-UP… — Attachment 9
Issue: In *In re: Social Media Adolescent Addiction/Personal Injury Products Liability Litigation*, Meta Platforms argues that five witnesses it designated for trial were properly and timely disclosed under the court's pretrial orders, and that the State Attorneys General lack grounds to strike them from Meta's trial witness list. The dispute turns on whether Meta's disclosures complied with applicable scheduling requirements and, if not, whether any deficiency was substantially justified under Fed. R. Civ. P. 37(c)(1). Meta further argues that the AGs' supplemental filing in support of their motion to strike was itself procedurally improper, having been filed without leave of court in violation of local rules governing motion practice.
Filed on June 4, 2026, by defense counsel Ashley M. Simonsen, this is a procedural notice submitted by Meta in the pretrial phase of MDL No. 3047 in the Northern District of California. The filing attaches twenty exhibits (Exhibits A–T) and a supporting declaration by Ms. Simonsen, submitted to complete the evidentiary record in support of Meta's earlier opposition (ECF 3060-1) to the State AGs' motion to strike five named witnesses from Meta's trial witness list (ECF 3056). Meta argues that the AGs' subsequent supplemental filing (ECF 3104) was filed without court authorization and asks the court to strike it as procedurally defective. The exhibits appear to include correspondence and disclosure records bearing on the timeliness and propriety of Meta's witness designations, though the full content of those materials was not available for independent review.
This filing does not advance any substantive legal doctrine — it is pretrial housekeeping in one of the most consequential social media liability cases currently pending in federal court. Its practical stakes are nonetheless real: if Meta's five challenged witnesses survive the motion to strike, they will be available at trial to contest causation, damages, or platform-design claims before the jury in the State AG bellwether proceeding. How Judge Gonzalez Rogers resolves the dispute over the AGs' supplemental filing may also offer a signal about how strictly motion practice discipline will be enforced as this MDL moves into its trial phase. For those tracking the litigation, the document is less a source of new legal rules than an indicator of where the parties are expending effort in the final stretch before trial.
NOTICE by Meta Platforms, Inc. OF FILING OF FOLLOW-UP… — Attachment 7
Issue: In *In re: Social Media Adolescent Addiction/Personal Injury Products Liability Litigation*, Meta Platforms argues that five witnesses it intends to call at trial were properly disclosed to the State Attorneys General during fact discovery, making exclusion unwarranted. The dispute turns on whether Meta's disclosures were sufficiently specific to satisfy the MDL's discovery obligations and, separately, whether the AGs filed an unauthorized supplemental brief that the court should disregard.
At the pretrial stage of this MDL, the State AGs moved to strike five witnesses from Meta's trial witness list, contending that Meta failed to adequately disclose them during discovery (ECF 3056). Meta filed this Notice on June 4, 2026, as a follow-up to its opposition (ECF 3060-1), attaching a declaration from its counsel Ashley M. Simonsen and twenty exhibits — including what appear to be deposition notices, correspondence, and discovery materials — intended to document that the five witnesses were in fact disclosed with adequate notice. Meta simultaneously asks the court to strike the AGs' own supplemental filing (ECF 3104), characterizing it as submitted without leave of court and therefore procedurally improper. The Notice itself does not develop extended legal argument; the twenty exhibits constitute the operative substance of Meta's position.
This is a procedural housekeeping dispute internal to the MDL, but its practical stakes are real: losing witnesses on the eve of trial can meaningfully reshape a party's trial strategy and narrative. The outcome will turn on the specific language of this MDL's case management orders and whether Meta's documentary record is sufficient to show the AGs had genuine notice and opportunity to conduct discovery on the contested witnesses. The filing creates no new law and carries no significance beyond this litigation, but it illustrates the high-volume procedural friction that characterizes large MDL proceedings as they approach trial.
MOTION for Leave to File Brief Amici Curiae
Issue: In *In re: Social Media Adolescent Addiction/Personal Injury Products Liability Litigation*, the First Amendment Coalition and Free Law Project argue that a proposed court order cannot lawfully compel members of the general public—including non-party archival organizations—to delete and destroy copies of a court document that was briefly available on the public docket before being sealed. The question is non-obvious because it sits at the intersection of a district court's undisputed power to seal its own records and the independent constitutional and statutory rights of third parties who lawfully obtained those records during the window of public access.
At the pretrial stage of this major MDL, the parties submitted a joint stipulated proposed order (ECF No. 3048) that would, among other things, require deletion and destruction of copies of ECF No. 3009-1—a document briefly accessible on PACER before being sealed. Non-parties First Amendment Coalition and Free Law Project, represented by Professor Eugene Volokh and attorney Megan Gray, filed this motion for leave to submit an amicus brief opposing the delete-and-destroy provision. The proposed amicus brief advances three independent arguments against the order as applied to members of the public who lawfully obtained the document: that Rule 65 does not permit injunctive relief binding non-parties outside the narrow categories of agents and active participants; that ordering deletion without notice or a hearing violates basic due process guarantees; and that compelling removal of lawfully obtained public court records constitutes a prior restraint on speech in violation of the First Amendment. Free Law Project separately argues that, as an interactive computer service hosting user-uploaded documents through the RECAP plugin, it is shielded from compelled removal by 47 U.S.C. § 230. The brief seeks denial of the deletion provision solely as applied to non-parties who accessed the document while it was publicly available.
This filing presses a structural question about the outer limits of judicial power that has rarely been litigated directly: once a court document becomes publicly accessible—even briefly—can a federal court order ordinary people and legal-records archives to delete it? The answer has immediate stakes for the social media MDL, where internal corporate documents that appeared on PACER may be effectively irrecoverable from public archives, but the implications extend to every MDL and high-profile federal case where inadvertent public filing occurs. The convergence of prior restraint doctrine and Section 230 in a court-records context is novel, and if accepted even in part, could meaningfully constrain district courts' equitable reach over third-party hosts like CourtListener and analogous legal transparency infrastructure. Three open questions flagged here—whether inherent docket-management power extends to ordering destruction of records held by non-parties, whether Section 230 bars affirmative injunctive relief compelling content removal, and how Rule 65's "active concert" prong applies to systematic PACER-mirroring services—are likely to recur as litigation-document archives become more organized and publicly accessible.
Reply Brief — Attachment 2990
Issue: In *In re: Social Media Adolescent Addiction/Personal Injury Products Liability Litigation*, Meta Platforms and Instagram argue that an opposing damages expert should be excluded at the Daubert stage because his "Bad Experience Violations" methodology impermissibly counts harms arising from third-party content—conduct Meta contends Section 230 immunizes—and because his core extrapolation projects an 11-day, largely non-U.S. internal survey across six years without any statistical validation. The case raises the non-obvious question of whether Section 230 immunity can operate not merely as a defense to liability at the pleading or summary judgment stage, but as a freestanding basis to exclude an expert's quantification methodology under FRE 702.
Meta and Instagram filed this reply brief on April 24, 2026, in support of their pending motion to exclude or strike the expert opinions of Carl Saba, a damages witness offered by the California Attorney General and three co-plaintiff states in MDL No. 3047. The brief responds to the AGs' opposition and seeks full exclusion of four of Saba's opinions (Ops. 2–5) under FRE 702 and *Daubert v. Merrell Dow Pharmaceuticals*. Meta argues that Saba's "Bad Experience Violations" count necessarily attributes liability to Meta for third-party user content, placing the methodology beyond the reach of any cognizable legal theory under Section 230. Meta separately contends that Saba's central empirical foundation—an 11-day internal company survey called BEEF, predominantly reflecting non-U.S. data—cannot reliably support a six-year, nationwide harm projection under *General Electric Co. v. Joiner*, and that the Ninth Circuit's 2025 decision in *Engilis v. Monsanto Co.* forecloses the AGs' argument that such deficiencies go only to weight. Finally, Meta argues that Saba's time-spent thresholds were set by counsel rather than derived through any independent expert methodology, and that his disgorgement figure lacks both a causal nexus to specific wrongdoing and a statutory basis under the applicable consumer protection laws.
The Section 230 argument is the most doctrinally ambitious piece of this filing: if accepted, it would establish that Section 230 immunity can collapse the Daubert admissibility inquiry—barring an expert from quantifying harm attributable to third-party content even when the underlying claims have survived dismissal. That would mark a significant procedural extension of immunity doctrine well beyond its traditional deployment at the pleading stage, and courts in this MDL have already drawn lines that complicate Meta's position. The BEEF-survey extrapolation challenge is the brief's strongest technical argument, representing a clean application of *Joiner*'s analytical-leap standard to a fact pattern—counsel-selected, geographically limited, temporally narrow survey data projected across years—that is difficult to rehabilitate through rebuttal alone. More broadly, this filing is worth watching because the expert exclusion fight will shape what the jury-facing damages case looks like in one of the first state AG consumer protection trials to proceed in this MDL, and a successful Daubert challenge here could effectively cap the states' ability to quantify violations at scale.
EXHIBITS re 2480 Brief, Partially Unsealed Meta Exhibits…
Issue: In *In re: Social Media Adolescent Addiction/Personal Injury Products Liability Litigation*, the Plaintiffs' Steering Committee argues that Meta and co-defendants designed their platforms with features — including infinite scroll, autoplay, algorithmic notification timing, and gamification mechanics — that were unreasonably dangerous for minor users, and that commercially feasible safer alternatives existed and were knowingly bypassed. The legal question is whether that evidence is sufficient to create genuine disputes of material fact on design defect and feasibility, precluding summary judgment in Defendants' favor on core products liability claims.
Filed on April 13, 2026, as Amended Exhibit 989 to the Plaintiffs' Omnibus Opposition to Defendants' Motions for Summary Judgment (Dkt. 2480), this exhibit is the expert report of Tim Estes, dated May 16, 2025. Estes, who founded the child-safety platform AngelQ, offers opinions that Defendants' platforms were defectively designed through deliberate deployment of compulsive-engagement mechanics targeting minors, relying in part on Defendants' own internal documents to establish contemporaneous knowledge of foreseeable harm. He further argues that meaningful age verification and parental control mechanisms — including credit card checks, government ID scanning, and federated identity systems — were commercially available and in use by comparable platforms such as Xbox, Apple, and Google Family Link well before Defendants implemented them. The report contends that Defendants' eventual parental controls were opt-in, structurally ineffective, and arrived only after harm was already documented, characterizing their inadequacy not as a missed opportunity but as itself a design defect. Estes cites COPPA, the 2023 U.S. Surgeon General's Advisory, and the Kids Online Safety Act as reinforcing the regulatory and normative baseline against which Defendants' design choices should be measured.
This report represents a significant moment in the effort to establish that products liability design defect doctrine applies to social media platform architecture — a theory that, if credited at summary judgment, would move the litigation past the threshold question of legal viability and into full merits adjudication. The feasibility argument is particularly consequential: by grounding safer alternative design in real-world commercial comparators that predated the alleged harm period, Plaintiffs aim to foreclose any claim of technological impossibility as a matter of law, converting feasibility into a jury question. Two open doctrinal questions hang over the report's reception: whether courts will apply a minor-specific risk-utility standard for engagement features that serve adult users while foreseeably harming children, and whether COPPA compliance functions as a regulatory floor or a safe harbor that displaces common law claims — neither of which has been definitively resolved in this MDL. The report's individual causation gap and its use of Estes's own platform as a feasibility comparator are predictable pressure points that Defendants will likely press in both Daubert proceedings and in reply briefing.
STATEMENT OF RECENT DECISION pursuant to Civil Local… — Attachment 2940
Issue: In *In re: Social Media Adolescent Addiction/Personal Injury Products Liability Litigation*, the State Attorneys General of California, Colorado, Kentucky, and New Jersey argue that Section 230 of the Communications Decency Act does not immunize Meta from state consumer protection claims premised on Meta's own deceptive conduct and product design choices — the central defense Meta is pressing in its pending Motion for Summary Judgment. The question is non-obvious because Section 230(e)(3) expressly preempts inconsistent state laws, and federal courts, including the Ninth Circuit, have historically construed that immunity broadly against claims that would treat platforms as publishers of third-party content.
Filed on April 13, 2026 — two days before oral argument on Meta's Motion for Summary Judgment — this is a Statement of Recent Decision submitted under Civil Local Rule 7-3(d)(2), a procedural mechanism that allows a party to bring a newly issued judicial opinion to the court's attention without extended argument. The filing attaches as Exhibit A the Massachusetts Supreme Judicial Court's April 10, 2026 decision in *Commonwealth v. Meta Platforms, Inc.*, No. SJC-13747, which the State AGs characterize as holding that Section 230 does not bar state consumer protection claims against Meta. The notice does not brief the underlying legal arguments; it presents the Massachusetts ruling as directly on-point authority supporting the AGs' opposition to Meta's summary judgment motion. The filing makes no attempt to reconcile the Massachusetts holding with controlling Ninth Circuit precedent, and it does not address whether the Massachusetts consumer protection statute is materially similar to the statutes at issue in this MDL.
Meta's core defense in this MDL is that Section 230 shields it from state liability for harms caused by its platforms — a defense that, if accepted at summary judgment, could end the case before trial. The State AGs are pointing to a brand-new ruling from Massachusetts's highest court as evidence that courts are increasingly unwilling to let Section 230 block consumer protection claims about how Meta designed and marketed its products, and the eve-of-argument timing is plainly strategic. Whether the filing moves the needle depends entirely on whether the MDL court finds the Massachusetts reasoning persuasive under Ninth Circuit law — a question this notice conspicuously declines to answer. More broadly, the filing adds one more data point to an emerging question in the courts: whether state attorneys general suing in their sovereign enforcement capacity occupy a distinct doctrinal position under Section 230 that is not yet resolved by existing federal precedent.
STATEMENT OF RECENT DECISION pursuant to Civil Local…
Issue: Commonwealth v. Meta Platforms, Inc.* asks whether Section 230 of the Communications Decency Act immunizes Meta from state-law claims targeting the deliberate design of addiction-inducing platform features — infinite scroll, autoplay, and intermittent-reward notifications — engineered to exploit adolescent neurology, and from claims based on Meta's own affirmative misrepresentations about Instagram's safety. The question is non-obvious because Meta argued these design features are inseparable from its role in curating and amplifying third-party content, which courts including the MDL district court had previously accepted as a basis for immunity.
The Massachusetts Supreme Judicial Court, on direct appellate review of an interlocutory order, unanimously affirmed the Superior Court's denial of Meta's motion to dismiss on Section 230 grounds. The SJC held that Section 230 immunity requires satisfying two elements: a dissemination element (the claim must be premised on the defendant circulating third-party content to others) and a content element (liability must turn on the specific harmful substance of that third-party content). The Commonwealth's design-feature claims failed the content element because addictive architecture operates identically regardless of what any user posts — it is content-indifferent — meaning neither element was met. Claims based on Meta's own marketing misrepresentations and deliberate failures of age-verification were held categorically outside Section 230 because the statute addresses liability for another's content, not a defendant's own speech. The Court explicitly criticized the MDL district court's contrary 2023 and 2024 rulings as failing to engage with Section 230's common-law foundations or congressional purpose. Plaintiff States filed the opinion in this MDL on April 13, 2026, as a Statement of Recent Decision in connection with pending dispositive motion Doc. 2779.
This opinion, from the highest court of Massachusetts, establishes the most analytically rigorous framework to date for limiting Section 230 immunity in platform-design cases, grounding a formal two-element test in a careful reconstruction of common-law publisher liability that competing courts will find difficult to dismiss as result-oriented. It directly and by name repudiates the MDL district court's Section 230 rulings, creating an explicit record of contrary authority as the Ninth Circuit considers an appeal of those very rulings argued in January 2026. For the AG plaintiffs in this MDL, the opinion supplies both doctrinal ammunition — a ready-made analytical framework — and a high-court imprimatur for the proposition that content-indifferent design claims fall entirely outside Section 230's scope. The Court left open whether the design-defect framing alone would independently defeat immunity and flagged without deciding that Meta's push-notification system may render Meta an information content provider, preserving additional avenues for future plaintiffs.
JOINT CASE MANAGEMENT STATEMENT for April Case… — Attachment 2
Issue: In *In re: Social Media Adolescent Addiction/Personal Injury Products Liability Litigation*, defendants Meta and YouTube argue that Section 230 of the Communications Decency Act immunizes virtually every platform feature plaintiffs allege caused harm to adolescents — including recommendation algorithms, autoplay, infinite scroll, and engagement-maximizing notifications — on the theory that these constitute protected "publishing" decisions over third-party content rather than independent product design choices. The instruction also asserts that failure-to-warn claims are equally immunized, treating a platform's silence about its own design-generated harms as equivalent to an editorial decision about user-generated content — a position no circuit has cleanly endorsed.
At the federal MDL pretrial stage in the Northern District of California, the Plaintiffs' Steering Committee filed a Joint Case Management Statement for the April 2026 Case Management Conference, attaching as Exhibit 2 a redline of Defendants' Revised Proposed Jury Instruction #18, titled "Protection for Publishing and Expressive Activity." The instruction is defendants' revised attempt to secure court adoption of a broad Section 230 immunity charge after a prior version was rejected at the March 18, 2026 Pretrial Conference (ECF No. 2837). The proposed instruction enumerates sixteen categories of protected platform conduct and structures the only viable path to liability — involving undefined "Non-Protected Conduct," a negligence finding, and a substantial-factor causation showing — as a compounding standard that must be satisfied simultaneously. A case-specific passage further instructs jurors to consider evidence of third-party bullying solely to explain a plaintiff's continued platform use, preemptively limiting how causation arguments based on harmful content exposure may be presented. No case citations appear within the instruction text; the Section 230 framework is invoked implicitly through the instruction's definitional structure.
The platforms are asking the court to tell jurors, as a settled legal matter, that nearly everything plaintiffs challenge — recommendation algorithms, autoplay, infinite scroll, engagement notifications — is legally protected activity that cannot give rise to liability, effectively resolving the most contested open question in Section 230 law inside a jury trial rather than through a dispositive motion. The Supreme Court's 2023 *Gonzalez v. Google* decision deliberately left unresolved whether algorithmic amplification constitutes "publishing," meaning whatever the court decides about this instruction could become the most significant judicial statement on that question to emerge from this MDL. The court's prior rejection of an earlier version signals meaningful skepticism, and if the court issues a written ruling explaining why it again rejects or substantially rewrites the instruction, that order — not the instruction itself — may carry the greatest precedential weight for how future social media injury plaintiffs are permitted to frame their claims.
PRETRIAL ORDER NO. 2 RE: MOTIONS IN LIMINE by Judge… — Attachment 2898
Issue: This pretrial order addresses whether defendants in a social media products liability MDL may introduce expert testimony characterizing adolescent harm claims as unsupported or overstated, and whether plaintiffs may rely on lay witness accounts of exposure to inappropriate content as circumstantial evidence of harm without closing the causation loop through expert testimony alone. The questions are non-obvious because they sit at the intersection of FRE 702 gatekeeping, the design-defect theory that has so far allowed these claims to survive § 230, and the practical reality that exclusion of defense experts at the bellwether stage locks in the plaintiff's causal narrative for the jury. The stakes are amplified by the MDL's bellwether structure, in which evidentiary rulings here will likely shape how parallel cases are tried or settled across hundreds of consolidated actions.
Judge Yvonne Gonzalez Rogers issued Pretrial Order No. 2 on March 30, 2026, resolving ten motions in limine filed by both sides in advance of the bellwether trial in Breathitt County School District v. Meta Platforms Inc. et al. The court excluded both of defendants' affirmative experts: Dr. Hutt was barred for offering impermissible legal-conclusion advocacy rather than methodologically grounded opinion, and Dr. Hampton was excluded for circular, ipse dixit reasoning underlying his "no evidence of harm" and "moral panic" framing. Plaintiff's core defect theories — CSAM-reporting failures and content-filter labeling defects — survived defendants' challenge and will be presented to the jury, while plaintiff's lay testimony regarding exposure to inappropriate images was held sufficient to support a permissible inferential harm argument. The court denied both sides' overbroad motions on content moderation evidence and indicated a limiting instruction would follow, denied most sealing requests, and deferred one ruling on financial mismanagement evidence pending submission of audit materials.
The categorical exclusion of both defense experts creates a materially asymmetric evidentiary posture at trial: defendants enter without credentialed methodological opposition to the foundational claim that social media causes adolescent harm, while plaintiffs' specific design-defect theories proceed intact. The court's acceptance of circumstantial lay testimony as sufficient to support an inferential harm argument is a notable departure from the more demanding causation standards applied in other complex products liability contexts — such as pharmaceutical MDLs — and may prove contentious on appeal or in parallel proceedings where defense experts have survived Daubert scrutiny. The circumscribed admission of foreign regulatory evidence bearing on defendants' knowledge and feasible alternative design opens a significant avenue for plaintiffs across the MDL to introduce EU and UK regulatory findings without triggering foreign-law instructions, and the deferred financial mismanagement ruling leaves open a question that could bear directly on punitive damages framing in downstream bellwether cases.
Exhibit List DEFENDANTS PRELIMINARY EXHIBIT LIST… — Attachment 2851
Issue: Insufficient text to determine. (The document is a pretrial exhibit list, not an opinion, order, or brief addressing a specific legal question under §230, First Amendment doctrine, or AI/product liability theory.)
Defendants Meta Platforms, Inc. and Instagram, LLC filed a preliminary exhibit list on March 16, 2026, in the Breathitt County Board of Education bellwether trial within the MDL. The list discloses 235+ exhibits Defendants may use at trial, including Breathitt County School District financial statements, school board minutes, student handbooks, behavior and discipline data, technology plans, digital citizenship curricula, and COVID-related school communications. The filing notes that the list excludes impeachment or rebuttal documents and deposition transcripts, and that a final updated list is due April 20, 2026.
As a pretrial exhibit list rather than a ruling or substantive motion, this document does not advance legal doctrine; however, the categories of exhibits—particularly school financial records, pre-existing behavioral data, and district technology and digital-citizenship plans—signal that Defendants intend to contest causation and damages by attributing student mental-health and behavioral issues to pre-existing institutional, socioeconomic, and pandemic-related factors rather than to platform design.
Witness List by Meta Platforms, Inc. DEFENDANTS… — Attachment 2856
Issue: Whether Meta, Snap, TikTok, and YouTube are civilly liable under product design defect and related theories for harm allegedly caused to adolescent users of the Breathitt County School District, where defendants contest both causation and the adequacy of plaintiff's abatement damages model.
In this bellwether trial within the In re Social Media Adolescent Addiction MDL, defendants filed a corrected preliminary witness list on March 17, 2026, pursuant to a pretrial scheduling order, disclosing witnesses they may call live or by deposition designation in their case in chief at the Breathitt County trial. The list identifies Breathitt County school employees — several of whom served as Rule 30(b)(6) designees — whom defendants intend to examine on alternative causes of harm and the district's failure to mitigate, as well as platform employees from Meta, Snap, TikTok, and YouTube who will testify about each platform's safety policies, well-being tools, and internal research. Defendants also disclosed a substantial expert roster spanning psychiatry, epidemiology, education policy, marketing, economics, and platform design to challenge plaintiff's causation theory, damages model, and abatement cost estimates.
This witness list signals that defendants' trial strategy will center on contesting general and specific causation through scientific experts while affirmatively presenting evidence of platform safety efforts, positioning the case as a significant test of whether product liability theories can survive against social media platforms when defendants offer robust alternative-cause and reasonable-design defenses in the school-district plaintiff context.
Witness List by Plaintiffs' Steering Committee… — Attachment 2848
Issue: Whether Meta Platforms, TikTok/ByteDance, Google/YouTube, and Snap are civilly liable under a products liability theory — including defective design and failure to warn — for harms suffered by a school district caused by the addictive and engagement-optimizing design features of their social media platforms.
This is a plaintiff's preliminary trial witness list filed by the Breathitt County Board of Education in the Social Media Adolescent Addiction MDL before Judge Yvonne Gonzalez Rogers in the Northern District of California. The school district plaintiff designated 47 witnesses — including lay witnesses such as school administrators, counselors, and finance personnel; fact witnesses consisting of current and former employees of Meta, TikTok/ByteDance, Google/YouTube, and Snap; and expert witnesses spanning addiction medicine, neuroscience, computer science, public health, economics, and forensic accounting. The witnesses are expected to support the district's claims by addressing platform design choices prioritizing engagement over safety, algorithmic recommendation systems, failure to implement age verification, failure to warn of addictive platform features, quantified costs and diverted staff time the district incurred, and, if reached, punitive damages through financial testimony on defendants' wealth and ability to pay.
This witness list signals that the school district bellwether trial in the Social Media MDL is advancing toward trial on a products liability theory that characterizes engagement-optimizing algorithms and addictive design features as actionable defects — a framing that, if successful, could establish a roadmap for institutional plaintiffs to recover costs attributable to platform design independent of Section 230 immunity arguments previously litigated in the MDL.
ORDER GRANTING IN PART AND DENYING IN PART RULE 702… — Attachment 2857
Issue: Whether expert general causation opinions offered to show that specific platform design features cause compulsive use and mental health harms in adolescents must be excluded under Federal Rule of Evidence 702 because the experts fail to disentangle actionable design defects from content and conduct immunized by Section 230 and the First Amendment.
In this MDL products liability action brought by school districts and state attorneys general against Meta, Google, ByteDance, and Snapchat, defendants moved under Rule 702 to exclude thirteen plaintiffs' general causation experts on multiple grounds, including methodological unreliability, lack of qualifications, improper reliance on internal company documents, and—critically—failure to isolate the causal effects of design features the court had previously deemed actionable from features and content barred by Section 230 or the First Amendment. The court denied the Section 230/First Amendment exclusion argument across all twelve experts to whom it applied, holding that expert testimony need not independently establish every element of the plaintiff's case to be admissible, and that the experts' reports did in fact address the specific actionable design defects identified at the motion-to-dismiss stage. The court granted the motions in part, however, striking discrete opinions in which experts—lacking relevant expertise—opined that defendants consciously prioritized profit and engagement over user wellbeing, finding those opinions went beyond each witness's disclosed area of expertise.
This ruling advances the theory that product-design claims targeting social media platforms' compulsive-use-inducing features can survive both Section 230 immunity and First Amendment limits at the expert-admissibility stage, so long as expert opinions are tethered to the specific design defects the court has deemed actionable rather than to third-party content or protected publishing decisions—a framework that could shape how plaintiffs structure expert testimony in future platform-liability litigation.
Exhibit List PLAINTIFF'S PRELIMINARY EXHIBIT LIST…
Issue: Whether Meta's internal research, design decisions, and communications regarding adolescent users' mental health and well-being are admissible at trial to support plaintiffs' product liability claims arising from alleged addiction-causing features of Facebook and Instagram.
This document is Plaintiff's Preliminary Trial Exhibit List filed on March 16, 2026, in the multidistrict litigation consolidating personal injury claims against Meta and other social media defendants. The list identifies hundreds of proposed trial exhibits drawn from Meta's internal documents, including PowerPoint presentations, emails, internal research studies, and message summaries, covering topics such as teen mental health, problematic use research, social comparison effects, suicide and self-injury content, and Meta's internal awareness of risks to adolescent users. The exhibit list also includes congressional hearing transcripts, coroner reports, and video excerpts of statements by Meta executives and early investors. No court ruling is reflected in this document; it is a pretrial filing identifying evidence plaintiffs intend to introduce at trial.
The breadth and specificity of the exhibit list signals that plaintiffs intend to prove at trial that Meta possessed extensive internal knowledge of harms its platforms caused to adolescent users, which could be significant for establishing the knowledge and design-defect elements of product liability claims that courts in this MDL have allowed to proceed notwithstanding Section 230 immunity arguments.
Proposed Jury Instructions — Attachment 2837
Issue: Whether §230 of the Communications Decency Act and the First Amendment immunize social media platform defendants from liability for specific algorithmic and design features—such as infinite scroll, content recommendation algorithms, notification clustering, and autoplay—when a school district plaintiff alleges those features caused compulsive platform use and resultant mental health harms to its students.
The Breathitt County School District brought negligence and public nuisance claims against Meta, Snap, TikTok, and YouTube in MDL No. 3047, alleging that the defendants designed their platforms to foster compulsive use in minors, causing the district to expend significant resources addressing the effects on its schools. With trial set for June 15, 2026, the parties filed competing proposed jury instructions, with the central dispute concentrated in Instruction No. 18, where defendants proposed a detailed "Protected Conduct" list—encompassing algorithmic recommendations, infinite scroll, notifications, autoplay, and similar features—that the jury would be barred from using as a basis for liability under §230 and the First Amendment, while plaintiff objected and proposed a narrower instruction. The defendants' proposed instruction reflects prior court rulings categorizing specific platform features as protected, and confines plaintiff's recoverable claims to a limited "Non-Protected Conduct" list including age verification processes, parental controls, account deletion processes, and appearance-altering filters.
This document is significant because it reveals how §230 and First Amendment protections will be operationalized at the jury instruction level in the first bellwether trial of a major social media addiction MDL, effectively showing which platform design features a court has already ruled immune from tort liability; the outcome could establish a concrete, feature-by-feature framework for distinguishing actionable product design claims from immunized publishing decisions that other courts and litigants could adopt or contest in future platform liability litigation.
MOTION to Exclude and/or Strike Expert Testimony of Carl… — Attachment 2845
Issue: Whether expert testimony calculating statutory penalty "violations" under state consumer protection laws by counting teen users' encounters with third-party "bad experiences" on Instagram must be excluded under Federal Rule of Evidence 702 where that methodology treats Meta as a publisher of third-party content immunized under 47 U.S.C. § 230(c)(1).
In this MDL consolidating state attorneys general consumer protection claims against Meta, defendants moved under Federal Rule of Evidence 702 to exclude the testimony of plaintiffs' damages expert Carl Saba, a financial consultant who calculated statutory violations using two primary methods: extrapolating results of an 11-day internal Meta survey (the "BEEF Survey") over six years to count teen encounters with third-party harmful content, and counting instances of teen usage exceeding 30 minutes per day per month. Meta argued that Saba's "Bad Experience Violations" opinion is legally impermissible under Section 230 because it attributes liability to Meta for third-party content that Meta would have had to actively vet to avoid, that both violation-counting methodologies are incorrect as a matter of law under the four lead states' consumer protection statutes (which require violations to be tied to wrongful acts or exposed/injured consumers), and that Saba's methodology was in large part designed by counsel rather than grounded in his own expert analysis. Meta further sought exclusion of Saba's disgorgement opinions as lacking the required causal nexus between Meta's advertising profits and the alleged misconduct, and as impermissible under California, Kentucky, and New Jersey law, which permit only restitutionary disgorgement or none at all.
The motion presents a significant question about whether Section 230 immunity can be invoked not only to defeat substantive liability claims but also to exclude expert damages methodologies that treat a platform's publication of third-party content as the predicate "violation" for penalty calculation purposes, potentially extending §230's reach into the evidentiary phase of litigation. If the court grants exclusion on this ground, it would signal that plaintiffs in platform-liability cases must carefully disaggregate algorithmic and design conduct from publishing conduct even at the damages-quantification stage.
Reply Brief — Attachment 2839
Issue: Whether §230 of the Communications Decency Act and the First Amendment require exclusion, under FRE 403, of evidence concerning third-party content and protected publishing features (including recommendation algorithms, autoplay, notifications, and Snap Streaks) that plaintiff seeks to admit at trial to support failure-to-warn and product-defect claims against social media platforms for alleged adolescent addiction injuries.
In this MDL bellwether trial (Breathitt County Board of Education v. Meta et al.), defendants Snap, Meta, YouTube/Google, TikTok, and related entities filed a motion in limine seeking to exclude evidence of third-party content and §230-protected platform features; this document is defendants' reply brief in support of that motion, filed March 11, 2026, ahead of a March 18 hearing. Defendants argue that plaintiff's failure-to-warn theory impermissibly attempts to premise liability on protected publishing functions, contrary to the Ninth Circuit's holdings in *Doe v. Grindr* (9th Cir. 2025) and *Estate of Bride v. Yolo Technologies* (9th Cir. 2024), which require that any actionable harm be "independent of the site's publishing function." Defendants further contend that plaintiff's claimed "actionable features" rationale for admitting the challenged evidence is pretextual, and that the evidence presents a serious risk of unfair prejudice under FRE 403 because a jury would likely draw a direct causal inference from protected conduct rather than from any non-protected design defect.
This reply brief illustrates how the §230 immunity question is migrating from the pleadings and summary judgment stages into trial-management rulings, testing whether the court's prior "feature-by-feature" liability framework can be operationalized as an evidentiary filter; the outcome could establish a replicable in limine standard for separating protected editorial/publishing conduct from actionable product-design claims in platform-liability litigation.
Administrative Motion to File Under Seal - Plaintiff's… — Attachment 3
Issue: Whether Section 230 and the First Amendment bar admission of evidence regarding YouTube's and Meta's platform design features—including "likes," recommendations, age verification, autoplay, shorts, notifications, and negative third-party content—in a minor plaintiff's negligent design trial premised on addiction rather than third-party content.
In this California state court MDL bellwether trial (JCCP 5255), Google/YouTube moved in limine to exclude all evidence of allegedly negligent design features on Section 230 and First Amendment grounds, relying in part on the Ninth Circuit's 2025 decision in *NetChoice v. Bonta*, which applied strict scrutiny to a statute restricting platforms from displaying like-counts to minors. Judge Kuhl denied the motion in its entirety, holding that where plaintiffs' theory is that content-agnostic design features (such as the "like" mechanism, notifications, autoplay, and data-driven algorithmic profiling) made the platform addictive independent of any specific third-party content, neither Section 230 nor the First Amendment shields defendants from liability or bars the underlying evidence. The court further held that even features for which liability is precluded under Section 230—such as age verification defaults—remain admissible as contextual evidence explaining why other design choices were allegedly negligent, and that negative third-party content is admissible to illustrate the degree of a plaintiff's addiction rather than as a basis for publisher liability.
This ruling advances a significant and recurring distinction in platform liability litigation: that Section 230 and the First Amendment operate as liability bars tied to *content-based* claims, not as blanket evidentiary shields against design-defect theories premised on addiction-inducing, content-agnostic features, potentially signaling that state-court juries will hear extensive evidence about algorithmic architecture even where direct liability for that architecture is nominally cabined by prior rulings.
Administrative Motion to File Under Seal - Plaintiff's… — Attachment 2830
Issue: Whether evidence of content and platform features is subject to exclusion at trial under §230 of the Communications Decency Act and the First Amendment in a products liability action brought by a school board against Meta and other social media defendants.
This is a temporary sealing motion filed by Plaintiff Breathitt County Board of Education on March 9, 2026, in the MDL proceeding before the Northern District of California, seeking to seal its opposition to Defendants' Motion in Limine #1, which moves to exclude evidence of content and features allegedly protected by §230 and the First Amendment. The substance of the underlying opposition brief is not included in the filed document, as the motion relates solely to the administrative sealing request. No ruling on the merits of the evidentiary dispute is reflected in this filing.
Insufficient text to determine the precise arguments or the court's reasoning, but the existence of a motion in limine framing §230 and the First Amendment as evidentiary shields — rather than pleading-stage defenses — signals that defendants are pursuing these protections through trial to limit what a jury may consider regarding platform content and design features.