Chinese language language agency DeepSeek’s new AI chatbot superior the positions of the Beijing authorities 60 % of the time in response to prompts about Chinese language language, Russian, and Iranian false claims, a NewsGuard audit found.
…
NewsGuard examined DeepSeek with a sampling of 15 Misinformation Fingerprints, NewsGuard’s proprietary database of falsehoods throughout the data and their debunks. The sampling included 5 Chinese language language false claims, 5 Russian false claims, and 5 Iranian false claims. […]
Adjust to the most recent data and protection debates on sustainable agriculture, biomedicine, and completely different ‘disruptive’ enhancements. Subscribe to our publication.
For its audits of AI devices, NewsGuard applies three speedy sorts that replicate how prospects make use of generative AI fashions for data and information: “innocent,” “most important,” and “malign actor.” (For additional information on NewsGuard’s red-teaming methodology, speedy and rating definitions, and former month-to-month AI audits, see proper right here.)
NewsGuard’s analysis of how DeepSeek responds to prompts related to disinformation from China, Russia, and Iran found that the chatbot repeated false claims even in response to neutral, straightforward queries.
That’s an excerpt. Be taught the distinctive put up proper right here