Chinese AI startup darling DeepSeek has been told to fix chatbot hallucinations or face a multi-million dollar fine.
Claim 70% Off TipRanks Premium
- Unlock hedge fund-level data and powerful investing tools for smarter, sharper decisions
- Stay ahead of the market with the latest news and analysis and maximize your portfolio's potential
Clearer User Warnings
Italy’s Competition Authority, AGCM, said after a lengthy probe that the Chinese firm had committed to launching a country-specific version of its chatbot in Italy, and make its warnings about chatbot hallucinations clearer and accessible to users.
It said DeepSeek will now “make its disclosures about the risk of hallucinations more transparent, intelligible and immediate.”
DeepSeek also pledged to lower the hallucination rate of its models through technical changes and update its terms and conditions and warnings in Italian.
The company must provide a progress report to regulators within 120 days or face a fine of up to €10 million ($11.7 million). The AGCM is still assessing whether DeepSeek’s services fall under the EU’s Digital Services Act, which could bring further regulatory requirements.
Hallucinations are Growing
The Italian authorities began investigating DeepSeek in June 2025 over concerns about insufficient user warnings.
According to Google (GOOGL), AI hallucinations are incorrect or misleading results that AI models sometimes generate. These errors can be caused by a variety of factors, including insufficient training data, incorrect assumptions made by the model, or biases in the data used to train the model.
A report by NewsGuard in September revealed that hallucination rates for top AI chatbots had nearly doubled year-over-year, rising from about 18% in 2024 to 35% in 2025. Inflection, Perplexity, and Meta (META) were highlighted as having some of the highest error rates.
That could mean that other chatbot providers may face similar regulatory demands in the months ahead.
DeepSeek came to prominence last year when it rocked U.S. tech titans with a cheaper but just as effective AI model. But it has had its fair share of challenges since then. This included reportedly using banned Blackwell chips from U.S. semiconductor giant Nvidia (NVDA) to help it develop new models.
It is also a critical player in China’s attempts to grow its own AI sector and challenge U.S. dominance.
What are the Best AI Stocks to Buy Now?
We have rounded up the best AI stocks to buy now using our TipRanks comparison tool.
