> daily_signal(2026_05_19)

A federal jury rejected Elon Musk's OpenAI lawsuit in under two hours. The nonprofit question will not be decided on the merits.

PickBits Daily Signal · Tuesday, May 19, 2026

By Mark Pickering · 7 min read · May 19, 2026

// tl;dr

A federal jury in Oakland needed under two hours on Monday to throw out Elon Musk's lawsuit against OpenAI and Sam Altman. The nine-member advisory jury unanimously found Musk waited too long under the statute of limitations; Judge Yvonne Gonzalez Rogers adopted the verdict and dismissed the case after a three-week trial. The question Musk wanted answered (did OpenAI break its nonprofit promise) never reached the merits. Musk called it a "calendar technicality" on X and said he will appeal.
About two-thirds of US doctors are now using OpenEvidence to help make clinical decisions, and almost none of their patients are told. NBC News reporting (picked up across the wire on Monday) put the number at roughly 650,000 US physicians plus another 1.2 million internationally, used in nearly 27 million clinical encounters in April alone. About 60 percent of the searches are clinical-decision queries. The product is free for doctors and is paid for by pharma and medical-device advertising.
Tomorrow Meta begins an 8,000-person layoff, and no federal law requires it to disclose whether AI is the actual cause. US tech layoffs in 2026 crossed 113,863 workers across 179 events as of Monday, averaging 825 per day. The bipartisan AI Workforce PREPARE Act (S.3339) would force the disclosure. Oxford Economics found in January firms "don't appear to be replacing workers with AI on a significant scale"; Deutsche Bank named the pattern "AI redundancy washing"; Sam Altman acknowledged both phenomena in February.
Both parties in Congress now have a bill to keep kids off AI chatbots, and they do very different things. Senator Josh Hawley's GUARD Act (S.3062) passed Judiciary 22-0 on April 30 and bans AI companion chatbots for users under 18; the House version was introduced the same day by Reps. Blake Moore (R-UT) and Valerie Foushee (D-NC); Hawley last week publicly demanded Majority Leader John Thune call it to the floor. Senator Ted Cruz's CHATBOT Act (S.2714) does not ban; it requires "family accounts" and parental consent. Neither has been scheduled for a floor vote.

The week that was supposed to be about Google I/O opened with a federal jury verdict that closes one of the two big legal stories in AI. In Oakland on Monday, the case Elon Musk filed against Sam Altman and OpenAI in 2024 ended in under two hours without a single ruling on whether OpenAI actually broke a promise to stay a nonprofit. The jury found Musk waited too long to ask, and the judge adopted the verdict. Three other stories landing the same week explain why ordinary people should care that the question won't be answered. Two-thirds of US doctors are quietly using AI to help pick treatments; tomorrow Meta cuts 8,000 jobs and can say "AI did it" with no legal duty to prove it; and the same week Congress is moving two very different bills that decide whether your kid can talk to a chatbot at all. The legal hook on OpenAI is now off. The product hook on patients, workers, and kids is tightening.

A federal jury rejected Musk's OpenAI case in under two hours. Two-thirds of US doctors are using OpenEvidence behind their patients. Meta cuts 8,000 tomorrow with no AI-disclosure law on the books. Hawley wants to ban chatbots for kids; Cruz wants their parents called instead.

1. A federal jury rejected Elon Musk's OpenAI lawsuit in under two hours. The nonprofit question will not be decided on the merits.

On Monday, May 18 in Oakland, a nine-member federal advisory jury took less than two hours to unanimously find that Elon Musk waited too long under the statute of limitations to bring his claims against OpenAI and Sam Altman. U.S. District Judge Yvonne Gonzalez Rogers, of the U.S. District Court for the Northern District of California, adopted the advisory verdict and dismissed the case after a three-week trial. Musk filed in 2024 alleging that OpenAI and Altman violated a promise to keep the company a nonprofit; the suit asked the court to unwind OpenAI's nonprofit-to-for-profit conversion and remove Altman and co-founder Greg Brockman. None of that will happen. The case never reached a ruling on whether OpenAI actually broke a nonprofit promise; it ended on calendar grounds.

Musk posted on X that the outcome was a "calendar technicality" and said he will appeal. He has other live litigation against OpenAI on different theories (including a separate antitrust complaint), but the headline case (the one most often cited as the legal lever to force structural changes at OpenAI) is over at the trial-court level. The practical read: the for-profit OpenAI you talk to every day, the for-profit OpenAI that closed an $852B-valuation funding round in February and the for-profit OpenAI that the Pentagon and named US banks are buying from, is now the legal default. The only federal courtroom that was on track to publicly examine whether the conversion was lawful did not get to that question. If a different plaintiff brings a case on the merits, they will need standing (a separate hurdle) and they will be filing it well inside the statute window that Musk missed.

NPR coverage of jury dismissing Elon Musk's lawsuit against Sam Altman and OpenAI on statute of limitations — npr.org · May 18, 2026

Why this matters: If you use ChatGPT for work, for homework help, or for a hard conversation, the most public legal challenge that asked a court to unwind OpenAI's for-profit pivot just got dismissed without a merits ruling. The for-profit OpenAI is now the legal default unless someone else, with standing and a fresh calendar, files a new case. Action this week: open ChatGPT's terms of service (chatgpt.com/terms) and search the page for the words "non-profit" and "mission" so you know which version of the company you are clicking accept on; the conversion is now legally settled at this trial-court level. If your employer rolled out ChatGPT Enterprise this year, ask whoever signed the contract whether the deal references the nonprofit mission or the for-profit corporate parent OpenAI Group PBC; the answer matters for which entity is on the hook if something goes wrong. If you donated to OpenAI or to research adjacent to it before 2019, read the NPR piece linked below so you know what was and was not decided.

npr.org/2026/05/18/nx-s1-5822366/musk-altman-openai-jury-verdict-claims-dismissed

2. About two-thirds of US doctors are using an AI tool to help make clinical decisions. Few patients are told.

Consumer-facing AI in healthcare. NBC News spent two months speaking with more than two dozen doctors, hospital administrators, medical students, and healthcare researchers across the country (Hawaii to Maine) for an investigation published last week and widely picked up on Monday. The finding: roughly 650,000 US physicians (about two-thirds of all US doctors) are now using OpenEvidence, an AI-powered medical search tool, to help make clinical decisions. Another 1.2 million doctors use it internationally. The product was used in nearly 27 million clinical encounters in April alone. About 60 percent of all searches are clinical-decision queries: physicians describe a patient ("this profile, this condition, these comorbidities") and ask "what is the right treatment?" The tool returns a chatbot-style answer grounded in peer-reviewed medical literature.

Two design choices make it different from a general-purpose AI like ChatGPT. First, OpenEvidence is free for verified physicians; the bills are paid by pharmaceutical and medical-device advertising. Second, when the medical literature is genuinely inconclusive, the system declines to answer rather than hallucinate, a guardrail CEO Daniel Nadler has called the central design constraint of the product. Patients in NBC's reporting were mostly unaware their physicians used it; few hospitals require disclosure. OpenEvidence's user base nearly tripled from about 250,000 doctors in mid-2024 to today's 650,000, and the company has framed itself publicly as the replacement for UpToDate (the subscription clinical-reference standard physicians have used for decades, owned by Wolters Kluwer). The pattern matters because the room where your treatment is decided is now a room with an AI assistant, and patient consent and disclosure norms have not caught up.

NBC News investigation: two-thirds of US doctors quietly using OpenEvidence AI tool for clinical decisions — nbcnews.com · May 14, 2026

Why this matters: The next time you sit in a doctor's office and your treatment plan is explained to you, there is roughly a two-in-three chance the doctor consulted an AI tool to land on it. You probably will not be told. The tool is grounded in peer-reviewed literature and is built to decline when the evidence is thin, but you have no current right to ask whether it was used in your case. Action this week: for your next non-routine appointment, write down two questions to ask: "Did you use an AI tool like OpenEvidence to help with this recommendation?" and "What sources is the recommendation based on?" Both are fair questions and almost no patient asks them. If you manage a parent's care or a chronic condition in your household, open openevidence.com on your phone (it is patient-readable even though it is built for physicians) and search the condition by name before the appointment so you arrive with the same vocabulary your doctor will be using. If you work in IT or compliance at a hospital or clinic, pull up your AI-usage and consent policy and find out whether OpenEvidence is on the approved-tools list and whether disclosure to patients is required; the policy is usually two years behind the practice.

nbcnews.com/tech/tech-news/openevidence-ai-doctor-medical-physician-login-app-what-npi-uptodate

If a friend forwarded this to you and you want the signal in your inbox five mornings a week, subscribe at pickbitsai.substack.com. The daily is free.

3. Tomorrow Meta cuts 8,000 jobs. If they say AI did it, no federal law requires them to prove it.

Layoff disclosure gap. US technology companies have eliminated 113,863 workers across 179 distinct layoff events so far in 2026 (as of Monday, May 18), averaging 825 job losses per day. On Wednesday, May 20, Meta begins its 8,000-person reduction, the largest single layoff of the year so far; Reuters reporting frames it as the first installment of cuts that could eventually reach 20 percent of Meta's global workforce. Companies citing AI as the driver are doing so under no federal disclosure rule: under current law, an employer can publicly attribute a mass layoff to artificial intelligence and a laid-off worker has no legal right to verify the claim or see the underlying analysis. New York's amended WARN Act now requires AI-cause disclosure for layoffs in NY, but it is a state-only rule and most cuts fall outside it.

The structural skepticism is on the record. Oxford Economics concluded in a January briefing that firms "don't appear to be replacing workers with AI on a significant scale," and that the layoffs may be cover for routine cost-cutting. Deutsche Bank analysts named the pattern "AI redundancy washing" in a January note and said it would be "a significant feature of 2026." In February, asked directly, OpenAI CEO Sam Altman conceded both sides: there is real AI-driven displacement, and there is "some AI washing where people are blaming AI for layoffs they would otherwise do." The bipartisan AI Workforce PREPARE Act (S.3339), introduced last year, would require companies to specify when AI was a substantial factor in a mass layoff, name the AI systems used, estimate the percentage of jobs attributable to them, and describe any retraining attempted before the cut. The bill has not moved this session.

Tech Times report: 113000 tech layoffs in 2026 with no federal AI-cause disclosure law — techtimes.com · May 18, 2026

Why this matters: If your company announces an AI-attributed layoff this year, neither you, your coworkers, nor a reporter has a legal right to see the analysis behind the claim. The line "AI is doing the work now" can be the truth, or it can be the press-release framing for an unrelated cost cut, and from the outside the two look identical. Action this week: if you are at a company that has named AI in any recent cut, open the WARN Act notice your state requires (search "[your state] WARN Act lookup" and your state's labor department site) and read what the company actually filed; the gap between the press release and the WARN filing is informative. If you are in HR or people-ops, pull up S.3339 on congress.gov and skim the disclosure requirements so your team is not caught flat if a state copy-cat passes. If your role is at risk and your employer has invited you to retrain into AI-fluent work (the GM IT-skills profile from last week is the public-facing version), write down the three skills they are asking for and a 6-week study plan against them; the path back in is named, but the company will not draw the line for you.

techtimes.com/articles/316802/20260518/tech-layoffs-surpass-113000-2026-no-federal-law-requiring-ai-disclosure

4. Both parties in Congress now have a bill to keep kids off AI chatbots. They cannot agree on whether to ban it or just call your parents.

Federal regulatory action on AI companions for minors. Two competing federal bills landed in front of the same problem in the same five weeks. Senator Josh Hawley's GUARD Act (S.3062) passed the Senate Judiciary Committee 22-0 on April 30 and would ban AI companion chatbots for users under 18, require any AI chatbot to disclose its non-human status and lack of professional credentials, and add criminal penalties for companies whose chatbots solicit sexual content from minors or encourage self-harm. The bill now has 19 cosponsors (13 Democrats and 6 Republicans). On May 17, Hawley publicly demanded Majority Leader John Thune bring the bill to a floor vote, threatening to force the issue. The House companion was introduced the same day (April 30) by Rep. Blake Moore (R-UT) and Rep. Valerie Foushee (D-NC).

Senator Ted Cruz's CHATBOT Act (S.2714), introduced with Senators Brian Schatz (D-HI), John Curtis (R-UT), and Adam Schiff (D-CA), takes the opposite approach. It does not ban chatbots for minors; it requires AI chatbot companies to create family accounts for users under 13, obtain parental consent, default to high-safety settings, restrict manipulative engagement features, and prohibit targeted advertising to minors. Cruz's bill has 3 original cosponsors. Both bills moved out of headline-driving committee testimony about specific teen suicides linked to chatbot use, including the Character.AI / Sewell Setzer III case that anchored multiple hearings. Neither bill has yet been scheduled for a floor vote in either chamber. The two-track choice (full ban below 18 versus a parent-mediated permission system) is the substantive question Congress is now asked to answer this session, with midterm-cycle pressure rising on both sides.

Detroit News coverage of chatbot bills as midterms loom; Hawley and Cruz competing approaches — detroitnews.com · May 17, 2026

Why this matters: If you have a kid under 18 who uses ChatGPT, Character.AI, Replika, or any of the companion-style chatbots, the federal default could change this year, and the two bills moving fastest disagree on whether your kid is allowed to use one at all. The choice between Hawley's ban and Cruz's parental-consent system is not a small distinction; it changes whether a 14-year-old can log into a chatbot to do homework help, or whether you (as the parent) are the one clicking yes. Action this week: open congress.gov, search S.3062 and S.2714 and read the one-page bill summaries side by side so you know which is which when a school newsletter or PTA email links to a "AI for kids" bill; the two are not interchangeable. If your kid uses any companion chatbot today, sit down for ten minutes and look at the app together (Character.AI, Replika, Snapchat My AI, or the standard ChatGPT app), then write down which of the two bills you would actually want signed; the answer is the position your member of Congress should hear. If you teach or run a school district's tech program, pull up the National Association of Attorneys General letter from August 2025 that 44 AGs signed on chatbot harms to minors and save the URL; it is the bipartisan baseline most state copy-cat bills are now using.

detroitnews.com/story/news/politics/2026/05/17/chatbot-bills-children-minors-safety-hawley-cruz-free-speech

» What to watch this week

Google I/O 2026 keynote, today at 1pm ET. The keynote starts after this digest goes out. The four pieces to watch: the next Gemini model (rumors point at 3.2 Flash or a full version bump), the Aluminium OS / Googlebook laptop category formally launching with Acer, ASUS, Dell, HP, and Lenovo, the first Android XR glasses hardware (Samsung, Gentle Monster, and Warby Parker partners), and pricing on whether Gemini Intelligence stays a Pixel / Galaxy exclusive or rolls broader. Tomorrow's signal carries the actual recap.
Meta May 20 8,000-job cut takes effect. Reuters framing suggests Reality Labs absorbs the heaviest share; Bloomberg expects ads / social engineering to take a portion. The unit distribution by Wednesday afternoon is the single tell on whether this is a strategy reshape under new chief AI officer Alexandr Wang or a routine cost cut wrapped in AI framing.
Musk appeal filing on the OpenAI verdict. Musk said on X he will appeal. The Ninth Circuit notice of appeal window is 30 days. Whether his lawyers file (and on what theory: standing, equitable tolling, or new claim) is the indicator on whether the merits question is dead or just paused.
House floor schedule for GUARD Act companion. Speaker Johnson has not put the Moore-Foushee bill on the House calendar. A floor date (or Hawley forcing a Senate vote first) is the next visible milestone on whether either chatbot bill makes it past committee this session.
First named hospital system that requires patient disclosure when OpenEvidence is consulted. The pattern in NBC's reporting was near-zero disclosure. The first major health system to publish an AI-tool patient-disclosure policy that explicitly includes clinical-decision support is the watershed; ask your own provider whether the policy exists yet.
Connecticut SB 5 signing date. Governor Lamont's office has signaled the omnibus AI-and-online-safety bill will be signed. The signing-statement text on chatbot-for-minors sections will tell us whether Connecticut is endorsing a Hawley-style ban or a Cruz-style consent model at the state level.

Tomorrow's signal lands here.