According to a report by Reuters on Monday (June 16th), the Italian antitrust regulatory authority, the Autorità Garante della Concorrenza e del Mercato (AGCM), has launched an investigation into the Chinese artificial intelligence (AI) startup DeepSeek for allegedly failing to adequately warn users that its AI system may generate false information.
DeepSeek has yet to respond to Reuters’ email request for comment.
The Italian regulatory authority, which is also responsible for safeguarding consumer rights, stated in a release that DeepSeek failed to provide users with “sufficiently clear, immediate, and understandable” warnings about the risks of “hallucinations” in the content produced by its AI.
The term “hallucinations” refers to situations where “one or more responses generated by the AI model after certain commands are entered contain inaccurate, misleading, or fictitious information.”
In February this year, another Italian regulatory authority, the Data Protection Authority, ordered DeepSeek to block access to its chatbot because the company had not addressed concerns regarding privacy policies.
The regulatory authorities questioned DeepSeek about its use of personal data, particularly seeking information on what personal data was collected, from which sources, for what purposes, based on what legal basis, and whether it was stored in China.
There are concerns from the public regarding DeepSeek and its potential ability to generate and spread fake news or misinformation.
Pro-China stance: DeepSeek’s AI responses often align with the Chinese government’s position, potentially suppressing or distorting information that contradicts the official Chinese narrative.
Incorrect and false information: Studies have shown that DeepSeek’s chatbot has a low accuracy rate, often repeating false claims and failing to provide reliable information, causing concerns about its dissemination of misinformation.
Chinese government influence and surveillance: DeepSeek originating from China raises concerns about its obligation to share data with the Chinese government for large-scale data collection and surveillance.
Information manipulation and censorship: DeepSeek has been found to censor or alter politically sensitive content, potentially influencing public opinion and spreading narratives supportive of the Chinese government.
NewsGuard review: The anti-misinformation organization NewsGuard found that in 83% of cases, DeepSeek failed to accurately convey news and current affairs content.
US Congressional report: A report by the US House of Representatives’ Select Committee on US-China Strategic Competition stated that DeepSeek poses a national security threat and accused it of data theft and manipulation for political purposes.
Cyabra research: Research by the online analytics company Cyabra indicates that DeepSeek has a coordinated network of fake accounts online, suspected to have involvement from the Chinese government to boost its visibility and influence.