Yazılar

Italy Probes Chinese AI Firm DeepSeek Over Misinformation Risks

/in /tarafından

Italy’s antitrust and consumer protection agency AGCM announced Monday it has launched a formal investigation into Chinese artificial intelligence startup DeepSeek, alleging the company failed to clearly warn users about the potential for its chatbot to generate false or misleading information.

The regulator stated that DeepSeek’s platform does not provide “sufficiently clear, immediate and intelligible” alerts about the risk of AI-generated “hallucinations” — a term used in the AI field to describe instances when models produce inaccurate or completely fabricated information in response to user prompts.

AGCM is focusing on the consumer rights aspect, emphasizing the risk users might unknowingly rely on erroneous AI outputs due to insufficient warning or transparency.

DeepSeek did not immediately respond to requests for comment on the investigation.

This marks the second run-in DeepSeek has had with Italian authorities this year. In February, the country’s data protection regulator ordered the startup to suspend access to its chatbot within Italy after the company failed to resolve concerns related to its privacy policy.

The probe highlights increasing regulatory scrutiny over generative AI models in Europe, particularly regarding transparency, data protection, and consumer rights.

France’s Mistral Launches Europe’s First AI Reasoning Model to Challenge US and China

/in /tarafından

French AI startup Mistral has unveiled Europe’s first AI reasoning model, aiming to rival leading American and Chinese competitors by leveraging logical thinking for complex problem-solving. The launch marks a significant step in Europe’s bid to carve out a homegrown presence in the competitive AI landscape.

Mistral’s reasoning models utilize “chain-of-thought” techniques, enabling the AI to generate intermediate reasoning steps when tackling difficult questions. This approach could help overcome current limitations faced by the industry’s traditional strategy of simply scaling up model size with more data and computing power.

Backed by venture capital at a $6.2 billion valuation, Mistral differentiates itself by emphasizing its European roots and commitment to open source, contrasting with proprietary models from companies like OpenAI and Google. French President Emmanuel Macron has publicly supported the startup, highlighting its strategic importance.

Mistral’s product lineup includes an open-source Magistral Small model available for free download on Hugging Face, and a more advanced Magistral Medium tailored for business clients. The models support reasoning in multiple languages including English, French, Spanish, Arabic, and simplified Chinese.

While American AI giants have largely kept their most advanced reasoning models proprietary, Chinese firms like DeepSeek and Alibaba have adopted open-source approaches to showcase their technology. Meta has integrated reasoning capabilities into its latest models but has yet to release a standalone reasoning model.

Industry observers see Mistral’s launch as Europe’s best chance to catch up in the AI arms race, particularly as the field shifts focus from brute-force scaling to more sophisticated reasoning abilities.

Rednote Joins Wave of Chinese Firms Releasing Open-Source AI Models

/in /tarafından

Chinese social media platform Rednote (Xiaohongshu) has released an open-source large language model named dots.llm1, joining a growing number of Chinese tech companies making AI models publicly available. This open-source move contrasts with many U.S. tech giants like OpenAI and Google, which keep their most advanced AI models proprietary, although some American firms such as Meta have also embraced open-source AI.

The release aims to showcase China’s technological prowess, foster developer communities, and extend global influence amid U.S. export restrictions targeting China’s advanced semiconductor industry.

According to Rednote’s technical paper published last Friday on Hugging Face, dots.llm1 performs comparably on coding tasks to Alibaba’s Qwen 2.5 model but is less advanced than models like DeepSeek-V3.

Rednote, known for its Instagram-style platform where users share photos, videos, and text, ramped up AI development after OpenAI’s ChatGPT debut in late 2022. Recently, it launched Diandian, an AI-powered search app for its main platform.

Other Chinese companies following this open-source path include Alibaba, which introduced the upgraded Qwen 3 model in April, and startup DeepSeek, whose low-cost R1 model has made waves globally for its competitive performance despite lower development costs.