AI and Surveillance: The Invisible Danger

Table of Contents

Digital Freedom - This article is part of a series.

Part : This Article

We Know This Playbook
#

There is a script. We have seen it play out several times in recent years. A technology is presented as inevitable. Critics are marginalized. Regulators nod along. And by the time the public wakes up, the infrastructure is already built.

During Covid, the mantra was: “Trust the Science.” Measures were pushed through on an emergency basis. Anyone who asked questions was a conspiracy theorist. The EMA approved vaccines under fast-track procedures — under political pressure, not scientific rigor.

Now the same thing is happening with Artificial Intelligence. Only quieter. And more permanently.

The EU AI Act: Protection or Fig Leaf?
#

In February 2025, the first prohibitions of the EU AI Act came into force. Social scoring? Banned. Real-time biometric mass surveillance? Banned. Sounds good. Too good.

Because the exceptions are the real law. Law enforcement agencies are allowed to use real-time facial recognition — for “serious crimes” and with judicial authorization. Retrospective biometric identification is not banned, merely classified as “high-risk.” [1]

Biometric verification — the question “Are you who you claim to be?” — is completely exempt. That is not a loophole. That is an invitation. [2]

151 Million Euros Don’t Buy Pizza
#

The tech industry spends 151 million euros annually on lobbying in Brussels. An increase of over 50 percent in four years. Meta alone invests 10 million euros. Microsoft and Apple 7 million each. [3]

In the first half of 2025: 378 lobby meetings with senior Commission officials and MEPs. Three meetings per working day. Most discussed topic: Artificial Intelligence.

Over 9 million euros flow annually to think tanks like Bruegel, CEPS, and CERRE — which then publish “independent” studies supporting their funders’ positions. [4]

In January 2026, Corporate Europe Observatory showed how Big Tech shaped the weakening of European digital legislation, article by article. [5]

This is regulatory capture. Word for word the same pattern as with the pharmaceutical industry.

Palantir: The Silent Guest in German Police Offices
#

Palantir — co-founded by Peter Thiel, grown large on CIA contracts — has long since arrived in German police departments.

Bavaria has been using Palantir’s Gotham since 2024. Hesse since 2017. NRW since 2019. In March 2025, Baden-Wuerttemberg signed a contract worth 25 million euros. [6]

What Gotham does: It merges databases. Police records, civil registry data, communications data, social networks. It detects “patterns.” It predicts who might become a problem. This is not social scoring — officially. It operates under “threat prevention.” [7]

The difference between predictive policing and social scoring is a branding problem. Not a technical one.

Clearview AI: 100 Million in Fines, Zero Paid
#

Clearview AI has collected over 30 billion facial images from the internet. Without consent. European data protection authorities have imposed fines totaling over 100 million euros — France, Greece, Italy, the Netherlands, the UK.

Clearview has not paid a single cent. [8]

In October 2025, noyb filed criminal charges in Austria — because fines do not work when a company simply ignores them. [9]

Max Schrems: “Clearview AI seems to simply ignore EU fundamental rights.”

Deepfakes: The End of Provability
#

A study with 2,085 participants shows: Fabricated videos cause significant reputational damage — regardless of technical quality. Even cheap forgeries are effective. [10]

By 2025, up to 8 million deepfake videos on social media. During the 2025 Canadian election, deepfake clips surfaced that looked like genuine CBC news reports. [11]

We are approaching the point where video evidence means nothing. Not because everything is fabricated — but because everything could be. That is the real damage. It does not destroy truth. It destroys trust in truth.

Parallels to Covid: Same Script, New Actors
#

Fast-tracking instead of due diligence. During Covid, approval timelines shortened from years to months. With AI, compliance guidelines are drafted on fast-track basis. Rules for high-risk systems do not apply until August 2026.

Regulatory capture. Pharmaceutical companies sat on panels deciding on their own products. Tech companies have 378 meetings per half-year with EU decision-makers. A 2025 study identifies AI safety as “a field with enormous potential for regulatory capture.” [12]

“Trust the experts.” During Covid: Do not question. With AI: You do not understand the technology. Both times, the argument undermines democratic oversight.

Concentration of power. During Covid: Pfizer, Moderna, BioNTech. With AI: Google, Meta, Microsoft, OpenAI. The raw materials have changed. The structure has not.

AI as Counter-Public Tool
#

And this is where nuance matters. AI is not only a tool of power. It is also a tool of counter-power.

This blog is an example. Articles are researched, structured, and proofread with AI assistance. Not to replace human thinking — but to amplify it. A single author with AI support can conduct research that previously required an entire editorial team.

The same technology that Clearview uses to catalog faces can be used to expose Clearview’s machinations.

The question is not: AI yes or no? The question is: Who controls it?

What Needs to Happen Now
#

Transparency for lobby meetings — not just the number, but the content
Enforceable penalties — if 100 million in fines can be ignored, no fines exist
Independent oversight — no panels where the regulated write their own regulation
Open-source mandates for publicly deployed AI — if Palantir is paid with my tax money, I want to see the code
Digital literacy — not “How do I use ChatGPT” but “How do I recognize when AI is being used against me?”