> daily_signal(2026_06_06)

New York's legislature voted to freeze new data centers for a year. The Supreme Court said the FCC can fine your carrier for selling your location.

PickBits Daily Signal · Saturday, June 6, 2026

By Mark Pickering · 10 min read · June 6, 2026

// tl;dr

On June 4, the final day of its session, the New York State Legislature passed a one-year moratorium on new data centers of 20 megawatts or more, the Responsible Data Center Development Act. Both chambers approved it; the Senate vote was 43 to 17. If Governor Hochul signs, New York becomes the first state in the country to halt big new data-center builds statewide. She has not said whether she will.
The Supreme Court ruled 8 to 1 on June 4 that the FCC can fine wireless carriers for selling customers' real-time location data without consent. Chief Justice Roberts wrote the majority; Justice Clarence Thomas was the lone dissent. The ruling upholds roughly $104 million in fines against AT&T and Verizon, part of a nearly $200 million 2024 FCC action that also hit T-Mobile and Sprint.
UK Labour MP Jess Asato filed a High Court claim on June 3 against xAI, the company behind Grok, over non-consensual sexual images the tool generated of her. Her lawyers call it one of the first cases to test whether the company that designs an AI image generator, not just the person who prompts it, can be held liable. She is seeking damages, a court declaration, and an injunction.
Cambridge researchers reported that the first vaccine designed entirely by AI cleared a Phase 1 human safety trial, 39 volunteers with no serious adverse events. The AI built a "super-antigen" aimed at a whole family of coronaviruses. The honest caveat: this tested safety, not protection, and it did not yet produce strong neutralizing immunity. A Phase 2 trial is planned.

For a month the fight over the AI buildout has been a local story, settled town by town and county by county, while the bigger questions about who pays and who is liable went unanswered. This week the answers started arriving from people with the power to set them. A state legislature voted to stop new data centers. The Supreme Court told the phone companies they can be punished for selling where you are. A court in London agreed to weigh whether the maker of an AI tool, not just its user, is on the hook for what it produces. Three of today's four stories are the same shape: the venue where the limits on AI get drawn this week was a government, not a company. The fourth is the exception that earns its place, an AI that built something useful instead of something that watches you.

This week a state legislature voted to halt new data centers, the Supreme Court let the fines stand against carriers that sold your location, a British lawmaker took an AI image generator to the High Court, and a vaccine an AI designed cleared its first human safety trial.

1. New York's legislature voted to freeze new data centers statewide for a year.

The data-center fight that has run county by county all spring just reached the whole state.

On June 4, the final day of its session, the New York State Legislature passed a one-year moratorium on new large data center construction. Both chambers approved it, with the Senate voting 43 to 17, as part of an omnibus package called the Responsible Data Center Development Act (bills S10642 and A11560, sponsored by Senator Kristen Gonzalez). The pause applies to facilities with a peak power demand of 20 megawatts or more, which covers the warehouse-scale sites that AI is driving, and the bill defines a "data center" broadly as anything drawing 1 megawatt or more. If Governor Hochul signs it, New York would be the first state in the country to put a blanket hold on big new data-center builds, rather than leaving each fight to a city council or county board.

The bill is not only a pause. After the year is up, sites of 5 megawatts and larger would have to meet labor standards, including prevailing wage and US-made iron and steel, and a requirement to run on 90% renewable electricity by 2040. The contrast with the last month is the point. The data-center pushback has been a string of local losses and standoffs, a Utah county that took the vote away from residents, a Texas developer who sued his county for $100 million, a Mississippi bill that put the cost on ratepayers. This is the first time an entire state legislature voted to stop and write the rules first. Hochul has not said whether she will sign; she has previously suggested that where data centers get built should be left to local governments, which is exactly the approach the moratorium overrides.

Why this matters: If you live in New York anywhere near a proposed data center, your legislature just voted to stop new big ones for a year while the state decides who pays for the power and water they pull. Even if you are nowhere near New York, this is the template every other statehouse now has in front of it: until this week the only people who could say no to a data center were a local board that the developer could sue or out-wait, and a whole state just took that decision up a level. Action this week: If you are in New York and a data center is proposed near you, call Governor Hochul's office (the public line is 518-474-8390) and say whether you want her to sign or veto, because the bill is on her desk now. Anyone can check whether a data center is planned in their own community at the AI Now Institute's tracker (datacenters.ainowinstitute.org/local). If you plan infrastructure or facilities for work, write down the two numbers that will spread to other states, the 20-megawatt moratorium trigger and the 90% renewable by 2040 mandate, before you commit to a site that a future state rule could freeze.

adirondackexplorer.org: State lawmakers pass one-year moratorium on large data centers (June 5, 2026)
nysenate.gov: S10642, Responsible Data Center Development Act (bill text and vote record)
thehill.com: New York lawmakers pass first-in-nation data center moratorium (June 4, 2026)

2. The Supreme Court said the FCC can fine your carrier for selling your location.

A privacy fight that started in 2024 over carriers selling where you are just got a final answer.

On June 4, the Supreme Court ruled 8 to 1 that the FCC acted lawfully when it fined wireless carriers for selling access to customers' real-time location data without their consent. Chief Justice John Roberts wrote the majority opinion; Justice Clarence Thomas was the only dissent. The case grew out of a 2024 FCC action that fined the major carriers nearly $200 million after the data they sold ended up in the hands of location-tracking middlemen and, in documented cases, bounty hunters. The fines in front of the Court were the AT&T and Verizon pieces, about $57 million and $47 million, roughly $104 million combined; T-Mobile and Sprint were part of the original action but not this appeal.

The carriers did not really argue that selling location data was fine. They argued the FCC's process was unconstitutional, that letting the agency impose the penalty in-house denied them their Seventh Amendment right to a jury trial. One appeals court had agreed with AT&T, another had upheld Verizon's fine, and that split is what the Court resolved. Roberts' answer was procedural but consequential: an FCC penalty order does not itself force a company to pay, so if a carrier refuses, the agency has to sue in federal court, where a jury is available, which means the jury right is satisfied. Thomas dissented, arguing the orders are coercive in practice. The bottom line for everyone who carries a phone is that the agency keeps a real tool to punish a carrier that sells your movements.

Why this matters: If you pay AT&T, Verizon, or T-Mobile, the Supreme Court just affirmed that the agency which can punish them for selling your real-time location, where you physically are, minute to minute, gets to keep doing it. The carriers were not fighting over whether they sold the data; they were fighting over whether the FCC was allowed to fine them for it, and they lost. That keeps a live deterrent on the table, but it is worth knowing the data sale already happened to real people before any of this, which is the part the ruling does not undo. Action this week: Turn off the location-data sharing your carrier still does by default. Open your carrier's privacy settings (AT&T's is the "CPNI" and "Enhanced Relevant Advertising" toggles in the myAT&T app, Verizon's is "Custom Experience" in My Verizon, T-Mobile's is "Advertising & Analytics" in the T-Mobile app) and switch the sharing off. While you are there, opt out at the carrier's privacy dashboard, and if you want the data brokers gone too, file a removal at the major people-search sites.

thedesk.net: Supreme Court says FCC can fine telecoms over customer location data (June 4, 2026)
thehill.com: Supreme Court upholds FCC fines on Verizon, AT&T over location data (June 4, 2026)

PickBits Daily Signal is free. If this lands in your inbox every day and it is worth something to you, the best way to support it is to share it with someone who would read it. Subscribe at pickbitsai.substack.com.

3. A UK MP took xAI to the High Court over Grok deepfakes of her.

The first UK court test of whether the company that builds an AI image generator, not just the person who posts, is on the hook.

On June 3, UK Labour MP Jess Asato, who represents Lowestoft, filed a High Court claim against xAI, the company that builds Grok, over non-consensual sexual images the tool generated of her. The images included a fabricated bikini photo and an explicit video that her filing describes as depicting her being drugged and prepared for an assault. Asato had publicly criticized Grok earlier in the year, and frames her own case as one instance of a much wider pattern of AI "digital undressing" that she says is harming thousands of women and children. Prime Minister Keir Starmer publicly backed the action, calling the images disgusting.

What makes the case matter beyond one MP is the legal theory. Her lawyers describe it as one of the first claims to test liability for the design of an AI system, not the conduct of the anonymous person who typed a prompt. The claim rests on data-protection law and the tort of misuse of private information, and it asks the court for three things: damages, a formal declaration that the conduct was unlawful, and an injunction ordering xAI to stop and to build safeguards into the tool. Almost every deepfake fight so far has chased the individual who shared an image or the platform that hosted it. If this one succeeds, it points the liability at the company that made the machine that produces the images in the first place.

Why this matters: If someone has ever used an AI app to make a fake nude of a real person, a classmate, a coworker, a stranger, you, this case asks the question that has had no clear answer: does the company that built the tool have to pay, or only the anonymous person who typed the prompt? Until now the law has mostly chased the poster, who is often impossible to find, or the platform, which hides behind hosting rules. A ruling that puts the maker of the generator on the hook would change the incentive for every company shipping image tools, because the cost of building no guardrails would finally land on them. Action this week: If intimate images of you, real or AI-faked, are out there or you fear they could be, use StopNCII.org, a free service that creates a digital fingerprint of an image so platforms can block it without you ever uploading the picture itself; in the UK the Revenge Porn Helpline (revengepornhelpline.org.uk) will help get content taken down. If you have teenagers, it is worth showing them both links this week, because the tools that make this are a tap away and the takedown tools are the defense.

lbc.co.uk: MP Jess Asato takes legal action against xAI over Grok deepfake images (June 3, 2026)
awo.agency: First claim in the UK against Grok's non-consensual deepfakes (filing law firm, June 3, 2026)
itv.com: Labour MP sues xAI over fake bikini image (June 4, 2026)

4. A vaccine designed entirely by AI cleared its first human safety trial.

The constructive story of the week, with the honest caveat attached.

Cambridge researchers reported this week that the first vaccine designed entirely using AI has passed a Phase 1 human safety trial. The vaccine, pEVAC-PS, is a needle-free DNA vaccine built on a platform called DIOSynVax. Instead of targeting one virus, the researchers used AI to design a "super-antigen," a synthetic protein carrying genetic features shared across the whole sarbecovirus family, the group of coronaviruses that includes both SARS-CoV-2 and the original SARS. In a trial of 39 volunteers, the vaccine showed no significant safety concerns or serious adverse events, and the results were published in the Journal of Infection. The team calls it the first time a vaccine designed entirely by AI and computer simulation has been tested in people.

Here is the caveat, and it matters: this trial tested whether the vaccine is safe, not whether it works. The AI-built antigen did trigger antibodies that recognized several coronaviruses at once, but it did not yet produce strong, broadly neutralizing immunity, and the immune response the researchers measured was, in their words, modest and variable. A Phase 2 trial is planned to find out whether it actually protects. The underlying trial also ran from 2021 to 2023; what is new is the published result. None of that erases the milestone. The promise of AI in medicine has mostly been slideware and someday; this is a concrete case where an AI designed a real medical candidate, a human being received it, and it came back safe.

Why this matters: If you have wondered whether all of this AI does anything useful for ordinary life rather than just watching you or coming for your job, here is one of the clearest yes answers yet: an AI designed a vaccine meant to work against a whole family of coronaviruses, a human trial gave it to 39 people, and nobody was harmed. That is the rare AI story where the win lands on the side of the people who would use the thing rather than the people selling it. It is also a useful lesson in reading hype: the headline is real, and the honest version is that it is safe so far, not proven to protect, which is exactly the distinction a careful reader should hold onto. Action this week: Read the Cambridge write-up at cam.ac.uk ("New universal vaccine technology could protect us from future virus outbreaks") so you have one concrete, non-hype example of AI doing good to point to the next time the conversation is all doom. If you work anywhere near health or science, the thing to watch is the planned Phase 2 trial, because that is where "safe" either becomes "it works" or does not.

cam.ac.uk: New universal vaccine technology could protect us from future virus outbreaks (June 5, 2026)
sciencedaily.com: AI-designed coronavirus vaccine passes first human safety trial (Journal of Infection, June 2026)

» What to watch this week

Whether Governor Hochul signs or vetoes the New York data-center moratorium. The bill is first-in-nation only if she signs. Watch for a signing or veto, and for which other state picks up the 20-megawatt-plus-2040-renewable template next.
How the carriers respond now that the Supreme Court has upheld the FCC's location-data fines. The deterrent is real again, so watch whether AT&T, Verizon, and T-Mobile actually pay, and whether anyone in Congress moves to write the consent rule into law rather than leave it to an FCC enforcement action.
Whether the UK High Court lets Asato's design-liability claim proceed. The novel part is suing the maker of the AI tool, not the user. An early ruling that the claim can go forward would be the signal that AI image-generator liability is now a live legal question, and worth watching for a parallel attempt in a US court.
Whether the Cambridge team's Phase 2 efficacy trial gets funded and scheduled. Phase 1 proved the AI-designed vaccine is safe. Phase 2 is where it either shows it can actually protect against the coronavirus family or does not, and that is the result that decides whether this is a breakthrough or a near miss.

Tomorrow's signal lands here.