AI Companies Have Paid Over $3.5 Billion in Fines in Europe and the US Since 2022

Regulators and courts in Europe and the United States have imposed fines and settlements worth more than $3.5 billion on artificial intelligence companies since 2022. That is the finding of a new analysis by Surfshark, reported among others by Rzeczpospolita and Parkiet, which tallied ten high-profile proceedings against seven technology companies.

Surfshark's tally covers the period from 2022 to 2026 and shows the scale of fines growing year over year. The earliest proceedings targeted Clearview AI, a company that scraped faces from the internet to build a facial recognition system, the Dutch, Italian, French and Greek data protection authorities fined it a combined roughly $105 million. Clearview, however, ignored most of the European rulings, arguing that EU regulators have no jurisdiction over a company operating from outside Europe.

The Industry's Biggest Fines

The record fine remains Anthropic's $1.5 billion settlement from 2025, reached over the use of pirated book copies to train its language models. Close behind is Meta, which paid $1.4 billion after it was found to have collected users' biometric data for facial recognition systems without their consent. These two penalties account for most of the total sum compiled by Surfshark.

Next come Google, fined $291 million for training models on copyrighted content, and Apple, which paid $250 million in 2026 in a case over exaggerated marketing claims about the AI capabilities of its devices. It is the only case in the tally where the problem was not training data but misleading communication with customers.

Data Without Consent, a Recurring Pattern

Surfshark's analysis stresses that in nine of the ten cases described, the core issue was the use of personal data, including biometric data, copyrighted content, facial images and recordings of children's voices, without user consent or an adequate legal basis. Amazon paid $25 million for retaining voice recordings of children collected by its Alexa assistant and using them to train the system without the required parental consent.

OpenAI appears in the tally twice. In 2024, Italy's data protection authority fined the company $17 million for training ChatGPT without a legal basis and for failing to meet its information obligations toward users. That decision, however, was overturned by an Italian court in 2026, which Surfshark cites as an example of the legal uncertainty regulators still face as they try to keep pace with the speed of technological development.

This could just be the beginning, accountability is starting to catch up with innovation, and the industry will need to rethink both how it builds AI and how it sells it - Dr. Luis Costa, head of Surfshark's research team

Europe's Data Versus AI Dilemma

Reports by Rzeczpospolita and Parkiet, the first Polish outlets to cover the analysis, point to the dilemma facing European regulators. On one hand, strict enforcement of data protection strengthens citizens' privacy and forces companies to be more transparent about the origin of their training data. On the other hand, the rising cost of fines and the need to adapt processes to legal requirements raise the cost of developing new models, which could widen Europe's technological gap with the United States and China.

For Polish companies and institutions using models from OpenAI, Anthropic, Google or Meta, Surfshark's tally is a reminder that the legality of training data is not a theoretical issue. National and EU data protection authorities, including Poland's UODO (the Personal Data Protection Office), are increasingly drawing on enforcement precedents set by their Italian, French and Dutch counterparts, and the growing body of precedent makes it easier for them to build their own cases against AI providers.

Surfshark's tally does not include cases still pending before courts and regulators, including numerous class-action lawsuits against language model providers in the United States and investigations by the European Commission under the EU's data protection regulation and the new AI Act. Additional fines could therefore push the total significantly higher before the year is out.

Sources: Ponad 13 mld zł kar za rozwój AI w Europie (rp.pl), Ochrona danych albo wyścig AI (parkiet.com), Big Tech slapped with $3.5bn in fines for using your personal data to train AI (techradar.com), Tech companies hit with $3.5B in AI fines since 2022 (digitalinformationworld.com)