Kimberlee Williams, a grandmother in Oklahoma, was arrested on a warrant from Maryland.
She had never been to Maryland.
An AI facial recognition tool made the match.
She spent six months in jail.
She's one of 15 known people this has happened to in the US.
A man was arrested at home in front of his wife and young daughters.
The only evidence against him was a facial recognition match to grainy surveillance footage.
Prosecutors dropped the charges.
A match isn’t evidence.
It’s a suggestion that looks like evidence.
AI browser agents can read your inbox, your history, and your open tabs to act on your behalf.
Researchers tested five of them and found at least one privacy vulnerability in every single product.
You asked it to book a flight.
It had access to everything else too.
Call centers have been running AI that analyzes your pitch and tone to score your emotional state during the call.
The EU just classified it as high-risk, banning workplace emotion-detection outright.
You thought you were on hold.
You were a training signal.
A data broker owned by the airlines sold your flight records, name, itinerary, financial details, to DHS and the IRS for years.
It told the agencies not to say where the data came from.
Only after lawmakers and reporters pushed did the program finally shut down.
Blue Shield of California shared the sensitive health information of 4.7 million people with Google Ads.
Not hacked.
Not leaked.
Handed over, as part of how the systems were set up to work together.
Your diagnosis became an ad targeting input.
72 members of Congress demanded an investigation into ICE and DHS buying your location data without a warrant.
One of the tools in question tracks up to 500 million phones through ordinary app ads.
The letter was bipartisan.
The purchasing continued anyway.
Records from 500,000 UK Biobank participants showed up for sale on a Chinese e-commerce site this year.
No hack.
No breach.
Just a dataset, listed like any other product.
They donated their DNA for research.
It ended up in a shopping cart.
Only 6 states require a warrant before police can fly a drone over your backyard.
In the other 44, an AI-piloted drone with a thermal camera and a license plate reader can be in the air before a patrol car even leaves the lot.
No federal rule says otherwise.
California just made it so you can no longer sue a company yourself for violating the state's wiretapping law.
Only the Attorney General can bring the case now.
One office, deciding which violations of your privacy are worth prosecuting.
Yours probably isn't.
An AI hiring platform is being sued for building secret candidate reports without telling applicants.
Social media profiles. Location data. Device tracking. Cookie data.
All compiled into a "likelihood of success" score.
You applied for a job.
It built a file on you first.
Supermarkets are installing cameras that read your facial expressions for frustration and indecision while you shop.
To dodge biometric privacy laws, they don't save your identity.
Just your face, your mood, and exactly which display made you hesitate.
A surveillance system called Webloc harvests location data from ordinary ad requests, your weather app, your flashlight app, to track up to 500 million phones.
Buyers include ICE and the U.S. military.
You never installed spyware.
You just checked the forecast.
Kroger made $527 million in a single year selling shopper data collected through its loyalty card.
Buyers included tobacco companies and financial firms.
You scanned a card for a discount on eggs.
You became a lead for a cigarette company.
An AI governance platform exists for exactly this gap, registering and classifying AI systems before they ship quietly, not after a reporter finds them.
https://t.co/2TuDNgGWC6
WIRED found an unreleased facial recognition system, code-named NameTag, inside Meta's smart glasses companion app.
It converts a stranger's face, captured through your glasses, into a biometric signature.
Meta says it rejected facial recognition.
Its own code says otherwise.
This is the actual problem: a company can say "we rejected facial recognition" while a governance gap lets it ship quietly in code.
A public statement isn't an AI inventory.
Systems like this need to be registered and classified before shipping, not after a reporter finds them.
27+ states now require a government ID or a selfie to prove your age online.
Discord lost 70,000 of those ID photos in a breach.
A dating app leaked 72,000 selfies.
Sold as child safety.
Built: a database of faces tied to what you looked at.
Florida's Attorney General just sued Netflix.
The claim: harvesting data from minors to build a global advertising business, while telling subscribers something else.
Kids don't sign consent forms.
Their viewing habits became ad inventory anyway.
Sam Altman, the CEO of OpenAI, said it himself: there is no doctor-patient confidentiality when your "doctor" is ChatGPT.
People are still typing their worst moments into it every day.
The company that built the confessional just told you it isn't one.
Personal information shows up in 65% of shadow AI incidents.
Not the employee's data.
Yours.
Your medical note.
Your loan application.
Your customer file.
Pasted into a tool you never used, sent to a server you'll never know about.