> daily_signal(2026_06_07)

ICE is giving more than a thousand local police departments an app that checks your immigration status from a face scan. Florida sued OpenAI's CEO personally.

PickBits Daily Signal · Sunday, June 7, 2026

By Mark Pickering · 10 min read · June 7, 2026

// tl;dr

Internal DHS documents show ICE plans to give more than 1,200 local and state police agencies across 32 states an app that scans a person's face and checks their immigration status in real time. It runs against roughly 250 million records, needs no warrant, consent, or notice, and a DHS document says a person cannot decline to be scanned. Any face captured is kept up to 15 years, including US citizens swept in by mistake.
On June 1, Florida Attorney General James Uthmeier filed an 83-page complaint suing OpenAI and CEO Sam Altman personally, the first state to put an AI chief executive's name on a lawsuit. It treats ChatGPT as a defective product released with no age checks and ties the harm to a 16-year-old's suicide and a campus shooting. Other states are weighing the same move.
SAG-AFTRA members ratified a four-year contract on June 4 by 91.42 percent, with 12 AI provisions. Studios can use a synthetic performer only if it brings "significant additional value" over a real actor or that actor's digital replica, and they have to get consent, pay, and give notice before scanning a performer's likeness. It is the most detailed AI rulebook any union has won.
Waymo and B2U Storage Solutions signed a deal on June 4 to give retired robotaxi battery packs a second life as grid energy storage in California and Texas. Waymo says the aim is hundreds of megawatts of storage, and B2U already runs an 8-megawatt plant in California built from more than 1,300 used electric-car packs. The worn-out battery goes back to work for years before it is ever recycled.

The fight over who controls AI has spent the spring being settled in town halls and county boards. This week it moved up to the people with real power to set the terms, and the answers cut both ways. A federal agency is about to put a face scanner in the pocket of every local cop in 32 states. A state attorney general put an AI founder's own name on a lawsuit over a dead teenager. A union of performers wrote consent and payment into the contract before the studios could scan them. And a robotaxi company found a use for AI's exhaust that helps instead of watches. Three of today's four stories are about a power being handed out or reined in, and the question each one answers is the same: when AI gets a new capability, who decides where the limit goes.

This week the government handed a face scanner to more than a thousand police departments, a state attorney general put an AI chief executive's name on a lawsuit, Hollywood's actors locked AI limits into their contract, and a robotaxi company sent its dead batteries back to the grid.

1. ICE is about to put a face-scanning immigration check in 1,200 local police departments.

The surveillance fight that has run through Flock cameras and location data all spring just reached the squad car.

Internal DHS documents obtained by 404 Media show that ICE plans to give potentially more than 1,200 local and state law-enforcement agencies, spread across 32 states and two territories, access to a phone app that scans a person's face and checks their immigration status in real time. The tool is the ICE Task Force Module app, the local-police version of the face-scanning software ICE's own agents already carry in the field. An officer points a phone camera at someone's face, and the app matches it against roughly 250 million records pulled from DHS and State Department databases, including passport and visa photos, and returns a status. It is built to work during an ordinary street stop or traffic stop, not just at a border crossing.

The documents describe a system with almost no friction built in. The scan needs no warrant, no consent, and no notice to the person being scanned, and a separate DHS document obtained earlier states plainly that a person cannot decline to be scanned. Once a face enters the database through one of these stops, the government keeps the scan for up to 15 years, regardless of whether the person turns out to be a citizen or has any immigration issue at all. Because the match runs on a phone in the field against a database that already holds most Americans' passport photos, the people most likely to be swept in by a bad match are exactly the ones the system is not supposed to be about: US citizens who happen to look like someone in the file.

Why this matters: If a local police officer stops you on the street or pulls you over, this puts a tool in that officer's pocket that can scan your face and pull up an immigration-status read in seconds, and the documents say you cannot legally refuse the scan. Even if you are a citizen with nothing to hide, a wrong match can land your face in a federal database for 15 years, and the people most exposed to a bad match are the ones who look like someone already in the file. This is the kind of capability that used to take a warrant and a trip to a station; it is about to ride along in a thousand squad cars. Action this week: Read the Electronic Frontier Foundation's plain-English writeup of ICE face scanning at eff.org (search "EFF Mobile Fortify") so you know what the tool is and is not allowed to do. Then read the ACLU's "know your rights" page for police stops at aclu.org/know-your-rights, where the load-bearing facts are that you can ask "am I free to go," you do not have to answer questions about your immigration status, and you do not have to unlock your phone. If you do community or legal work, save the 404 Media list of the 32 states in the rollout so you can check whether your local department is on it.

404media.co: ICE's plan to let cops around the country scan faces to verify immigration status (June 5, 2026)
rawstory.com: ICE to give 1,200-plus local police departments a face-scanning immigration app (June 5, 2026)
eff.org: Rights organizations demand a halt to ICE handheld face recognition (background)

2. Florida sued OpenAI and Sam Altman personally over harm to minors.

The first time a state has put an AI chief executive's own name on a lawsuit over what the product did to people.

On June 1, Florida Attorney General James Uthmeier filed an 83-page complaint in Florida's 10th Judicial Circuit against OpenAI and, unusually, against CEO Sam Altman personally. The suit treats ChatGPT as a defective product and accuses the company of deceptive trade practices, negligence, and product-liability violations for releasing a tool to the public with no real age verification or gatekeeping for minors. State lawyers and outside observers describe it as the first state-led lawsuit of its kind, and the first attempt by any government to hold an AI company's chief executive personally liable for user harm. Florida's complaint does not stand on abstractions; it names specific deaths.

The harms the filing leans on include the suicide of 16-year-old Adam Raine, whose family has separately sued OpenAI; the suspect in the April 2025 Florida State University shooting, who is reported to have queried ChatGPT; and the killing of two University of South Florida graduate students. The legal theory is the consequential part. By framing a chatbot as a product that can be defective, Florida is trying to drag AI into the same liability world as a faulty car or a dangerous toy, where a company can be sued for what it shipped rather than only for what a user typed. The "first to name the CEO personally" framing is the attorney general's characterization, not a settled point of law, but if even part of the product-liability theory survives a motion to dismiss, other states are already weighing whether to follow.

Why this matters: If you have a teenager who talks to ChatGPT, a state just argued in open court that the chatbot is a defective product that should have checked their age and warned them, and that the founder himself bears responsibility. Whatever happens to this particular suit, the theory it is testing, that you can sue the maker of an AI tool the way you would sue a carmaker for a faulty brake, is the one that decides whether these companies face real consequences or only bad press. That is the difference between a settlement that changes how a product ships and a headline that changes nothing. Action this week: If a teen uses ChatGPT in your house, open ChatGPT's Settings and turn on the parental controls and age safeguards OpenAI rolled out this spring (the setup link is at chatgpt.com/parental-controls), which let you link a teen's account and limit sensitive content. If you are not sure whether they use it, ask, and put the 988 Suicide and Crisis Lifeline number in their phone while you are at it. If you work in policy or compliance, pull the complaint itself and read how the defective-product argument is built, because it is the template the next state will copy.

npr.org: Florida sues OpenAI and Sam Altman over alleged safety lapses (June 1, 2026)
techcrunch.com: Florida sues OpenAI, Sam Altman in first-of-its-kind lawsuit (June 1, 2026)
cbsnews.com: Florida AG sues OpenAI and Sam Altman, citing risks to children (June 2, 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. Hollywood's actors locked AI consent and pay into their contract.

The most detailed AI rulebook any union has put into a signed agreement, ratified by a landslide.

On June 4, SAG-AFTRA national members voted to ratify their four-year 2026 TV and Theatrical Agreement by 91.42 percent, locking in a set of AI rules that took two rounds of hard bargaining to win. The contract runs from July 1, 2026 through mid-2030 and, by the union's own count, carries 12 distinct AI provisions. The headline rule limits when a studio can replace a human: producers may use a synthetic performer only if it brings "significant additional value" compared to a live union actor or that actor's own digital replica, a standard backed by arbitration and damages if a studio breaks it. The phrase is not a slogan; it is contract language a panel can enforce.

The provisions an ordinary worker would recognize are the ones about consent and money. Before a studio can scan an actor's face or voice to make a digital replica, it has to get the performer's consent and pay for it, and a new clause requires notice when a studio licenses an actor's data to an outside company, the kind of arrangement that could feed a performer's likeness into an AI training set. None of this stops the technology. What it does is force the consent, the payment, and the disclosure to happen before the scan, instead of leaving a worker to discover after the fact that their face is now a company asset. That is why other unions, from video-game performers to musicians, are reading this contract as the template for their own next fight.

Why this matters: If your face, your voice, or your work could ever be scanned and reused by an AI, the most detailed rulebook yet for forcing a company to ask first, pay you, and tell you when it hands your likeness to someone else just got ratified by performers, and it reads as a template any worker can point to. The win that travels is not the Hollywood-specific part; it is the structure, consent before the scan, payment for the replica, notice when your data is licensed onward. That is the exact list of protections most workers do not have when an employer wants to clone a voice or a face for an internal tool. Action this week: Read SAG-AFTRA's plain-language summary of the AI provisions at sagaftra.org (search "2026 TV Theatrical AI provisions") so you can see the actual "significant additional value" and digital-replica-consent language. If you are in a union or about to sign any contract that touches your voice, image, or written work, copy the three-part structure, consent, compensation, and third-party-licensing notice, into your own asks before you sign. If you manage people, write down that a likeness scan now comes with a known consent-and-pay standard, because your staff will start expecting it.

variety.com: SAG-AFTRA members ratify contract with AI terms and pension merger (June 4, 2026)
deadline.com: SAG-AFTRA members approve AMPTP deal by 91.42 percent (June 4, 2026)

4. Waymo is sending its dead robotaxi batteries back to the grid.

The constructive story of the week: AI's worn-out hardware going back to work instead of to the landfill.

On June 4, Waymo and B2U Storage Solutions announced a signed strategic supply agreement to give retired robotaxi battery packs a second life as stationary grid energy storage. As Waymo retires the battery packs from its Jaguar I-PACE fleet, B2U will take the used packs and wire them into grid-connected storage installations in California and Texas, the same states where Waymo runs its cars. Waymo frames the ambition as "hundreds of megawatts" of storage capacity over time. That figure is a stated goal rather than a contracted number, but it sits on top of a real track record: B2U already operates an 8-megawatt solar-and-storage plant in Lancaster, California, built from more than 1,300 used electric-vehicle battery packs.

The reason this is worth a slot is that it answers a real objection. A car battery that has lost enough capacity to be retired from a vehicle is usually still holding 70 to 80 percent of its original charge, which is useless for fast acceleration but perfectly good for sitting in a field and storing solar power for the evening. Routing those packs into grid storage means a single battery does years of additional work, soaking up cheap daytime power and releasing it at peak, before it is ever sent to be recycled. It does not erase the cost of building the batteries in the first place, and the megawatt goal is intent, not a delivered number. But it is a concrete case of an AI company's physical exhaust being put to a use that lowers a bill and steadies a grid rather than watching someone or replacing them.

Why this matters: If you have ever been told that electric-car batteries just turn into toxic landfill when the car is done, here is the counter: a robotaxi company is sending its worn-out packs back to work storing power for the grid, where a battery that is too weak to drive a car can still run for years. That is the quiet, useful version of an AI story, the kind where the win lands on the side of the people who pay an electric bill rather than the company selling the service. It is also a clean way to read the next "battery waste" scare, because second-life storage is the answer the scare usually leaves out. Action this week: Read Waymo's announcement at waymo.com/blog ("A second life for EV batteries") so you have a concrete example to hand the next time someone says EVs are an environmental con. If you are weighing home solar and battery storage, ask your installer whether they offer second-life packs, which run cheaper than new ones, before you pay for a brand-new battery. If you follow energy or sustainability for work, note B2U as the operator turning this from a press release into a running plant.

waymo.com: A second life for EV batteries, clean energy for communities (June 4, 2026)
techcrunch.com: Waymo's spent robotaxi batteries will be used as grid storage (June 4, 2026)
electrek.co: Waymo's retired robotaxi batteries are heading back to work with B2U (June 4, 2026)

» What to watch this week

Whether any state legislature or court moves to limit the ICE face-scanning rollout before it ships to local police. The documents describe a plan, not a finished deployment. Watch for a lawsuit, a state law restricting local participation, or a department publicly opting out, and for whether the 32-state list grows.
Whether OpenAI moves to dismiss Florida's suit, and whether a second state files its own. The defective-product theory lives or dies on the first motion to dismiss. An early ruling that the claim can proceed would be the signal that AI product liability is now a live legal front, not a press event.
Which union adapts the SAG-AFTRA AI language next. Video-game performers, musicians, and writers are all bargaining. Watch for the "significant additional value" standard or the consent-pay-notice structure showing up in another contract, which is how a Hollywood clause becomes a labor-wide norm.
Whether Waymo and B2U name an actual project, site, or megawatt figure. Phase one is a signed supply agreement with a stated goal. The thing that turns "hundreds of megawatts" from ambition into fact is a named installation with a real capacity and an online date, so watch for the first one.

Tomorrow's signal lands here.