Instagram story viewer> @mading.thok> Posts
11.5K
followers
47
following
Privacy & Data Protection | How-To | AI & TechNews #Sanity
Disclaimer: Tools utilizing Ai assist with production. I'm REAL.
POSTS STORIES REELS TAGGED
Download All
A sitting senator just held up a Walmart patent, on live television, for a shopping cart that reads your heart rate. 

Last week's Senate hearing on AI surveillance pricing named Amazon, Kroger, Lyft, Staples, Target, and JetBlue, with real numbers: a 62-page customer data file, 55 Lyft riders charged 29 different prices for the same ride, and a $700 billion industry by 2030. 

Companies deny doing this. Their patents say otherwise. What's coming next, and what you can do now. 

#SurveillancePricing #Surveillance #Pricing #SenateHearing #Privacy #ScreenSanity #SenateHearing #Walmart #ConsumerRights #DataPrivacy #senjoshhawley by @mading.thok
1k
2 months ago
Download
A former FTC chair gave Congress this exact example: an airline charging you more because your browsing history shows you just had a death in the family and need to fly fast. 

That's not hypothetical. Delta Air Lines is currently rolling out AI-driven ticket pricing to 20% of its domestic flights. Senators sent Delta's CEO a formal letter demanding answers. 

Here's what's real, what Delta says, and 3 things to check before your next flight. 

#SurveillancePricing #DeltaAirlines #DataPrivacy #AIPrivacy #PrivacyTips #ScreenSanity #TravelTips #AITravel by @mading.thok
746
2 months ago
Download
Your phone is not listening to your conversations. The real tracking is more boring than that, and honestly, way harder to escape. 

The FTC has a real case on record: a data broker's customer proposed building a digital fence around specific people's homes, just to track where they went, for years. 

Here's how it actually works, and 3 settings that help. 

#digitalprivacy #adtracking #iphoneprivacy #screensanity #madingthok #techexplained #onlinesafety #databrokers #privacytips #ftc by @mading.thok
106
2 months ago
Download
13 cases solved. 382 searches of an ex-girlfriend. Same kind of system. Most documented abuse was caught by audits, which means the numbers may be measuring detection rather than frequency — and that should worry you more about places with no audit at all. A federal judge just ruled warrantless access to a month of plate history is an unconstitutional search. So where should the bar be?

#SurveillanceSeries #LicensePlateReaders #PrivacyDebate #FourthAmendment #MadingThok by @mading.thok
1
5 hours ago
Download
Would you give an AI agent its own account on your computer? That's the actual Windows design — separate account, its own desktop session, read and write access to six folders. Real upside: software that does things instead of suggesting them. Real downside, quoting Microsoft: malicious content in documents can override agent instructions, leading to data exfiltration or malware installation — and once enabled, it's enabled for every user on that device. Off by default, still experimental. Decide before the prompt shows up.

#AIPrivacy #Windows11 #AIAgents #DigitalPrivacy #MadingThok by @mading.thok
2
a day ago
Download
Target's policy says chat agents see what you type before you hit send. That's their words. Grade: F. What to do:

1. Enable Global Privacy Control — Target honours it
2. Target.com → Your Privacy Choices → opt out of sale and targeted ads
3. Verify your opt-out status on the same page
4. Know that leaving Circle deletes your Rewards permanently
5. Uninstalling the app doesn't clear your device ID

#IsThisAppSafe #Target #PrivacyGrade #DigitalPrivacy #MadingThok by @mading.thok
0
2 days ago
Download
Four months ago I covered a Connecticut law requiring stores to post a sign when facial recognition is running. It took effect October 1. What it requires: signage at every public entrance, a QR code or link to the store's facial recognition policy, and matching only against a database the business maintains itself. 

Why it matters: 
Walmart's own policy lists face geometry as collectable, and has an "inferences" category covering intelligence and aptitudes drawn from your shopping patterns. 

The part nobody's saying: 
Reporting indicates AG enforcement of the signage rule doesn't begin until February 2027, and this is one state only.

#SurveillanceSeries #FacialRecognition #Connecticut #DigitalPrivacy #MadingThok by @mading.thok
5
3 days ago
Download
Two gates will quietly stop iOS 27's new parental controls from doing anything. In order:

1. Update every Family Sharing device to version 27 first
2. Settings → your name → Family → confirm a Child Account
3. Turn on Ask to Browse for new websites
4. Set Time Allowances per category, not one daily limit
5. Confirm Communication Safety is on, now covering gore and violence too

#ParentAndKids #iOS27 #ScreenTime #ParentalControls #MadingThok by @mading.thok
0
5 days ago
Download
Rationale: Scene 5, the Alpha response, never gets cut regardless of length. The 2010 precedent is the first thing to go because it needs its own setup to mean anything.

Spoken version:

One family was sent video of their daughter in her bedroom, in pajamas, recorded through her school laptop's webcam. The monitoring followed students home by default. Parents had to opt out.

This is Alpha School. Forty to seventy five thousand dollars a year. And the pitch is real. Two hours of adaptive software a day, and students have tested in the top one to two percent nationally.

But the model runs on behavioral data. WIRED and 404 Media reported screen recording, mouse and keyboard tracking, eye tracking, and webcam monitoring that flags facial expressions.

Alpha disputes this, and says the characterizations are false.

So before you sign anything at any school, ask three questions. What does the software record. Who outside the school can see it. And what does it do after the laptop leaves the building.

Sources scene: keep, shortened to 5s given the six citations, same six, faster scroll. by @mading.thok
1
6 days ago
Download
The software detects people, not just plates. One camera held 1.6 million images in 21 days. What to ask locally:

1. Request the contract from your city council — it's a public record
2. Ask the retention period in writing
3. Ask who can search it
4. Ask about federal sharing
5. Ask whether a searchable audit log exists

#SurveillanceSeries #LicensePlateReaders #DigitalPrivacy #PublicRecords #MadingThok

**Facebook** by @mading.thok
0
7 days ago
Download
Texas rejected this AI school's charter 10 to 3. Three months later the state was helping put the same software into public classrooms.

Reported Oct 1 to 2 by ProPublica, the Texas Tribune and Houston Public Media. At least ten districts approached, at least three pilots running, including Houston ISD.

In this one: what the software collects, who is actually on the other end when a student asks for help, and why the monitoring is how the model works rather than an extra.

I argue the other side too. Mastery based pacing is legitimate, the test results are externally measured, and the research on AI tutoring is cautiously positive.

Six questions to ask any school running AI learning software:

1. What does it record → screen, keystrokes, camera, microphone, location?
2. Is any of it analysed for attention, emotion or facial expression?
3. Who reviews it, including anyone outside the district or the country?
4. What does the device do after it leaves the building?
5. How long are recordings kept, and how do you request deletion?
6. If it is a public school → ask to see the contract.

Careful note: the monitoring described is documented at the company's private campuses. I have not seen reporting establishing it runs in the Houston pilot. Alpha School strongly disputes these characterisations.

#AIinSchools #StudentPrivacy #Parenting #EdTech #MadingThok by @mading.thok
0
7 days ago
Download
404 Media reported in September that OpenAI pays hundreds of contractors to read full ChatGPT conversations and rate the answers. Accounts are anonymized, but what you typed isn't. The part worth knowing: turning off AI training and turning off human review are two different controls. Microsoft states plainly in its Copilot FAQ that an opt-out of human review is not available. And Google says Gemini chats a reviewer has already seen are kept up to three years and are not deleted when you delete your activity.

1. ChatGPT: Settings → Data Controls → Improve the model for everyone → Off
2. Gemini: Gemini app → profile picture → activity settings → Keep Activity → Off, and set auto-delete to 3 months
3. Claude: Settings → Privacy → Model improvement → Off
4. Copilot: opt out of training; there is no human-review opt-out
5. Don't paste anything into a consumer AI chat you wouldn't want a stranger to read

#AIPrivacy #ChatGPT #DataPrivacy #AISafety #MadingThok by @mading.thok
2
8 days ago
Download
×

Download all media on this page

Photos Videos
back to up