Italy’s Data Protection Authority (DPA) goes after DeepSeek

Italy’s safety authorities have set their sights on the wave-making Chinese AI company, DeepSeek after Euroconsumers made a complaint to the Italian Data Protection Authority (DPA).

Euroconsumers is a group of consumer organizations in Europe. Last year, it launched a successful case against Grok over how it used data to train its AI.  

According to reports, the complaint concerns how DeepSeek manages personal data under GDPR, a data protection law in Europe. 

The Italian DPA has now written to Hangzhou DeepSeek AI and Beijing DeepSeek AI asking for additional information. DPA intends to uncover what personal data is collected, where it comes from, and why it collects them.

Da Roma con AI Amore!

The Italian DPA announced that it has sent a request for information to DeepSeek, particularly in relation to the training of its AI tools. @guidoscorza pic.twitter.com/bHElVZDkBo

— Luis Montezuma | @luismontezuma@mstdn.social (@montezumachavez) January 28, 2025

This includes details about the information used to train their AI systems and the legal reasons for processing this data. The entity also wants more information about computers in China

That’s not the end. DPA wants information about the startup’s web scraping activities for both registered and unregistered users.

DPA expects a response within twenty days.

Why suspicion is high around DeepSeek 

One major suspicion is that DeepSeek was created and has a home in mainland China. According to its privacy policy, the AI startup gathers and keeps information and data in its own country. That does not sit right with Euroconsumers. 

Secondly, the AI firm says in its policy that it transfers data to China. This immediately turns the AI tool into a political weapon – or rather, a topic of contention in the rising global sector.

What’s worse? Minor protection. According to reports, Euroconsumers pointed out that DeepSeek does not provide information on how it defends or limits minors’ access to its services, including age verification and the management of minors’ data.

Notably, DeepSeek’s age policy states that it is not meant for users under 18. However, it doesn’t explain how it enforces the rule. It simply recommends that users aged 14 to 18 read the data policy with an adult.

At a press meeting held by the European Commission, Thomas Regnier, Commission Spokesperson for Tech Sovereignty, answered whether there are worries at the European level over DeepSeek related to security, privacy, and censorship.

He replied, “The services offered in Europe will respect our rules […] These are very early stages, I’m not talking about an investigation yet.”

Sam Altman’s OpenAI says it has evidence of foul play

OpenAI says it has found proof that the Chinese AI startup used its proprietary models to train its own open-source competitor. According to reports, the ChatGPT maker in San Francisco said that it has noticed some signs of “distillation.” The ChatGPT maker assumes it may be from DeepSeek.

Distillation is a technique used by developers to obtain better performance on smaller models by using outputs from larger, more capable ones. This allows them to achieve similar results on specific tasks at a much lower cost.

Sounds like DeepSeek.

The launch of its R1 thinking model caught the attention of Silicon Valley markets. The problem is that OpenAI was founded 10 years ago, has 4,500 employees, and has raised $6.6 billion in capital.

On the other hand, DeepSeek was founded less than 2 years ago, has 200 employees, and was developed for less than $10 million.

If this distillation complaint is true, then it means that DeepSeek’s engineers lied. They said they had used about 2,000 Nvidia H800 chips. The chips are less advanced than the most cutting-edge chips, to train its model. This would mean DeepSeek could be using its rival model’s data, which would be a breach of OpenAI’s terms of service.

Cryptopolitan Academy: FREE Web3 Resume Cheat Sheet - Download Now