France probes China DeepSeek AI over privacy risks.

France’s data protection watchdog, CNIL, has launched an inquiry into Chinese AI startup DeepSeek over potential privacy risks associated with its chatbot. The move follows DeepSeek’s recent claim that it trained its DeepSeek-V3 model using less than $6 million worth of Nvidia H800 chips, a significantly lower cost than its competitors.

CNIL’s AI department is currently analyzing the tool and will question DeepSeek to better understand its data handling and compliance with privacy regulations. Other European regulators, including Italy’s data authority and Ireland’s Data Protection Commission, have also requested details about the company’s data processing practices, particularly concerning European users.

The European Union has some of the world’s strictest data protection laws under the General Data Protection Regulation (GDPR), with potential fines of up to 4% of a company’s global revenue for breaches. Additionally, the EU’s newly adopted AI Act imposes stringent transparency requirements on high-risk AI models, with penalties reaching up to 7% of global turnover.

DeepSeek now faces mounting regulatory scrutiny in Europe as authorities aim to ensure compliance with privacy standards while assessing the risks posed by advanced AI systems.