The hearings: the mechanism, on the record
The lawsuits ask who pays for the harm. The Congressional hearings ask how the machine works, and answer it in the subjects' own words. Between 2023 and 2026, four hearings put the mechanism on the record: a former Meta safety engineer who says the company measured the harm and chose not to act, five platform CEOs answering under oath, the trial lawyer who showed a jury the companies' internal documents, and a bipartisan fight over what Congress should do next.
Where the lawsuits establish that courts are willing to find harm, the hearings establish the design choice underneath it: safety tools that are off by default, unmeasured, or shipped to answer criticism rather than to work. That is the same brake-integrity question the rest of this site asks, when you tell the feed to stop, does anything actually stop, answered by the people who built the feed and the lawmakers who questioned them.
About this page. These four accounts are distilled from the official record of each hearing, seeded from the project's evidence-tiered claim ledgers, where corrections land first (each ledger, with its full quote-bank, is published in full). Two of them (2023 and 2024) are built from the official Senate transcripts. The other two (2025 and 2026) are recent enough that no official transcript is published yet, so they are a flagged tier-down, built from the witnesses' written testimony rather than the live question-and-answer, and marked as such below. Every quotation is verbatim from the record and attributed to the speaker and their role. Quotes were chosen for what they reveal about the mechanism, not for the viral moment: the selection is deliberately mechanism-not-tribe, and lines cut for being theater are named in the ledgers. As of July 16, 2026.
Four hearings, one mechanism
| Hearing | The record | What it puts on the record |
|---|---|---|
| Nov 2023, Senate (Bejar) | official transcript | A former Meta safety engineer: the company measured the harm and chose not to act. |
| Jan 2024, Senate (five CEOs) | official transcript | Five platforms under oath; senators show the safety tools do not hold. |
| May 2026, Senate (the verdicts) | tier-down (written testimony) | The internal company documents a jury was shown. |
| Dec 2025, House (the bills) | tier-down (written testimony) | The genuine, unresolved fight over what Congress should do. |
November 2023: the fake brake, named from inside
Senate Judiciary Subcommittee on Privacy, Technology, and the Law. Witness: Arturo Bejar, a former Meta safety engineer. Official transcript, S.Hrg. 118-663.
On November 7, 2023, a Senate Judiciary subcommittee heard sworn testimony from Arturo Bejar, an engineer Meta twice hired to make its products safer for children. Backed by his own memos to Mark Zuckerberg and Adam Mosseri that were entered into the record, he testified that the company measured how often children were harmed, reported the numbers to the top, and repeatedly chose not to act.
His most load-bearing line describes the exact pattern this site is about, a control that looks like a brake but does not stop the car: the safety features Meta shipped were, in his words, "a safety feature in name only to placate the press and regulators." He described the company "creating its own homework," narrow definitions of harm that let it grade itself. He asked for a help button a child could press when a message hurt them; it was not built. And when he proposed the obvious test, "let's measure our help by whether it helped," that "was not adopted."
The most concrete receipt in the hearing is a feedback brake that exists for advertisers and is withheld from children: "You can take an ad and say that is sexually inappropriate. But there's no way for a child to do that when they get a message." Senators from both parties named the design, not the speech, as the problem. Senator Josh Hawley (R-MO) called the company's public safety statistic "designed to mislead"; Senator Richard Blumenthal (D-CT) put it plainly: "It's a product. It needs to be different. It has to change."
No platform testified, so the companies appear here only as quoted documents. Their live answers are the next hearing.
Full ledger: every tiered claim and all the verbatim quotes are in the Nov 2023 Bejar distillation.
January 2024: five platforms, and the tools do not hold
Senate Judiciary Committee. Witnesses: the CEOs of Discord, Meta, Snap, TikTok, and X. Official transcript, S.Hrg. 118-497.
On January 31, 2024, the CEOs of Discord, Meta, Snap, TikTok, and X testified together before the Senate Judiciary Committee; three of the five appeared only under subpoena. Across seven hours and both parties, one pattern held: the companies described extensive safety tools, and senators showed the tools do not hold.
The brake is optional: Meta's Mark Zuckerberg confirmed that teens can opt out of the restrictive default settings he had just described. The brake is unmeasured: asked whether Meta reports the total amount of harmful content or only the share it catches, he answered, "No. I believe we focus on prevalence." The brake reaches almost no one: Snap's Evan Spiegel could name his parental-tool adoption, "approximately 200,000 parents use Family Center," against roughly 20 million teen users, about one percent; Discord's Jason Citron, asked whether his teen-safety tool works, said, "I do not have a study off the top of my head." And the brake was defunded: pressed on an internal memo, Zuckerberg was questioned about a rejected request for dozens of safety engineers that would have cost about $50 million in a quarter Meta earned $9.2 billion.
Senator Jon Ossoff (D-GA) named the tension at the heart of it before reading Meta's own SEC filing aloud: "You have a fiduciary obligation to your shareholders to get kids to use your platform more." Senator Richard Durbin (D-IL) reduced it to a sentence, put to Discord: "if that were working, we wouldn't be here today." The companies split on the child-safety bills: Snap and X endorsed specific ones ("we strongly support the Kids' Online Safety Act," said Spiegel), while Meta, Discord, and TikTok declined to commit to any.
Full ledger: every tiered claim and all the verbatim quotes are in the Jan 2024 five-CEO distillation.
May 2026: the documents behind the mechanism
Senate Judiciary Subcommittee on Privacy, Technology, and the Law. Witnesses: the trial counsel and expert behind the recent verdicts, and two survivor parents.
Tier-down. This hearing is recent enough that no official transcript is published yet. This account is built from the witnesses' written testimony, not the live question-and-answer; every witness was an advocate for legislation, with no platform present to answer; and the verdicts it describes are not final. Treat it as the weaker-evidence tier it is.
On May 13, 2026, a Senate Judiciary subcommittee built a hearing around something that had not existed before: American jury verdicts against the companies. Rachel Lanier, the trial lawyer who won California's K.G.M. verdict against Meta and Google (a verdict now on appeal), walked the committee through the internal company documents she had put before the jury. Those documents are this hearing's distinctive evidence.
The brake was never the goal: a 2016 YouTube presentation, as she quoted it, stated the aim to be "an app that is...Addictive." The brake was declined until forced: a 2020 Meta email called the company's unenforced under-13 rule "just indefensible" and recorded that the product team "would not prioritize this proactive work until we're legally required to do it." And the company's own staff treated the safety tools as beside the point: a 2023 Meta employee chat, reacting to the state lawsuits, includes the line "nobody is asking us to build 30 tools." That chat is dated three months before Meta's CEO offered those more-than-30 tools as the answer to the Senate, in the January 2024 hearing above.
A law professor, Mary Graw Leary, named the legal cause and the limit of litigation in one argument: the platforms "face no deterrence for designing products without regard to safety," and even the headline penalties are small against the scale, "$375 million dollar and a $6 million dollar verdicts" against parent-company revenue that "exceeded $400 billion" and "$200 billion last year." Two parents who lost children asked Congress to pass the Kids Online Safety Act.
The litigation posture, including the appeals, is in the lawsuit case files.
Full ledger (tier-down): the claim ledger and quote-bank is at the May 2026 verdicts distillation.
December 2025: the fight over the fix
House Energy & Commerce, Subcommittee on Commerce, Manufacturing, and Trade. Witnesses: a safety advocate, a pro-regulation lawyer, an industry association, and a civil-liberties group.
Tier-down. Same caution as above: no official transcript yet, so this account is built from written testimony, not the live exchange. Unlike the others, this hearing was built to disagree, and the witnesses genuinely do.
On December 2, 2025, a House Energy & Commerce subcommittee held a legislative hearing on a package of child-safety bills, including the Kids Online Safety Act, COPPA 2.0, and Sammy's Law. Its four witnesses, a safety advocate, a pro-regulation lawyer, an industry association, and a civil-liberties group, all agreed the harm is real. They split, sharply, on the fix, and that split is the most useful thing on this page: it is where the brake-integrity argument meets its hardest objections.
The design-regulation case came from Joel Thayer of the Digital Progress Institute, who argued the First Amendment is being weaponized to "turn our bulwark for free expression into a sword to cut down laws they don't like," and that the Supreme Court "categorically rejected TikTok's argument that the mere regulation of an algorithm raises First Amendment scrutiny." The industry position, from Paul Lekas of the Software & Information Industry Association, accepted safety law but pushed its shape toward "regulating design or privacy conduct, rather than content," to survive constitutional scrutiny. And the sharpest objection came from the civil-liberties witness, Kate Ruane of the Center for Democracy & Technology, who named the same mechanism the other hearings did, "the surveillance capitalism business model," but named it to argue the opposite remedy: that the durable fix is comprehensive privacy law, that age-verification mandates "create significant privacy risks," and that "Protecting children includes protecting their right to express themselves online."
Read honestly, this hearing does not confirm the brake-integrity thesis. It maps where that thesis meets its most legitimate objections, constitutional and strategic, and those objections belong on the record next to the mechanism the other three establish. The First Amendment cases the witnesses fight over are the same ones engaged in the policy paper.
Full ledger (tier-down): the claim ledger and quote-bank is at the Dec 2025 legislative distillation.
What the four establish together
Read as a set, the four hearings describe one mechanism from four vantage points, and the shape of each source is different enough that agreement between them is worth something. A whistleblower names the fake brake from inside one company (2023). Five CEOs answer for the same pattern under oath, across five platforms (2024). The internal documents a jury was shown supply the design intent behind it, in the companies' own words (2026). And a balanced panel argues over what to do about it, objections included (2025).
The throughline is the one this site is built on: the harm is not any single post but the delivery loop, and the safety controls offered against it are too easy to ignore, off by default, unmeasured, or, in the internal record, never meant to work. That is the case for a standard that tests a brake by whether it stops the car. The full argument, with its honest limits, is in the policy paper; the test is written out at the scorecard; the litigation record is in the lawsuits.
Sources
Each account is seeded from the official record of the hearing. Transcripts and written testimony submitted to Congress are U.S. government works.
- Nov 7, 2023, Bejar (S.Hrg. 118-663): official Senate transcript, govinfo CHRG-118shrg60432 (primary); Senate Judiciary hearing page.
- Jan 31, 2024, five CEOs (S.Hrg. 118-497): official Senate transcript, govinfo CHRG-118shrg57444 (primary); Senate Judiciary hearing page.
- May 13, 2026, the verdicts (tier-down; no official transcript yet): witness written testimony via the Senate Judiciary hearing page (primary for the written statements).
- Dec 2, 2025, the bills (tier-down; no official transcript yet): witness written testimony via the House Energy & Commerce committee repository (primary for the written statements).
Cleaned third-party transcripts were used only as a reading aid and cross-check, never as the quoted source. Every quotation on this page is verbatim from the official record above and attributed; the two tier-down hearings will be rebuilt to the full standard when their official transcripts are published.
Provenance and methodology: this page is a companion to the lawsuits explainer, covering the Congressional record rather than the litigation. The case content is sourced ledger-first: each hearing has an evidence-tiered claim ledger with a verbatim, attributed quote-bank, verified as an exact substring of the official record; this page is seeded from those ledgers, curated and not a full render, and corrections land in the ledgers first; the full ledgers, with their complete quote-banks, are published at /distillations. Quotes are selected for mechanism, not for the tribal or viral moment, and the two tier-down hearings are flagged as the weaker-evidence artifacts they are until their official transcripts publish. AI-assisted throughout: the framing and judgment are the author's, the drafting was collaborative. Updated on dated passes, not live; the as-of line above is the honesty backstop.