DeepSeek
DeepSeek business and news from across the web.
Latest coverage
US authorities accuse Chinese AI companies of industrial-scale campaigns to copy American models
US authorities, including the NSA, CISA, and FBI, have accused Chinese AI companies like DeepSeek and Moonshot of conducting industrial-scale campaigns to copy American AI models. These companies are alleged to be 'distilling' advanced American AI models for their own use.
DeepSeek's AI models are about to cost four times more
DeepSeek has announced significant price increases for its V4 AI models, with costs expected to quadruple. This change impacts the pricing structure for their advanced artificial intelligence offerings.
AI model distillation, explained: the cheap way to inherit an expensive model’s brain
AI model distillation, also known as knowledge distillation, is a technique where a smaller "student" model learns from a larger "teacher" model. This process allows for the creation of cheaper, more efficient AI systems by inheriting the capabilities of expensive, pre-trained models. However, concerns exist regarding the potential loss of safety knowledge and the inheritance of biases during this distillation process.
I regret to report that the AIs are weebs, too
A study by researchers from the University of the Basque Country and Cardiff University found that large language models like Claude, Gemini, and DeepSeek exhibit a disproportionate bias towards Japanese culture when responding to open-ended cultural prompts. This bias is amplified through instruction tuning, leading to a homogenization of cultural perspectives in AI outputs, with the United States and Japan being the most frequently referenced regions.
AI Yi-Yi!
DeepSeek has announced a 75% price cut for its AI services, while Meta is expanding its Hyperion AI supercluster plans to 5GW and increasing its Louisiana investment to over $50 billion. These moves indicate an acceleration in the competitive AI race.
AI Yi-Yi!
The article discusses the rapid advancement of AI voice cloning technology, noting that it can clone voices in minutes. It highlights concerns about intellectual property rights, using Taylor Swift's efforts to protect her voice as an example, and questions the rights of individuals regarding their AI-cloned voices. Additionally, it briefly mentions DeepSeek's development of its own AI chips.
Reuters: DeepSeek is developing its own AI chips
Chinese company DeepSeek, known for developing affordable AI models, is reportedly planning to develop its own AI chips. This move aims to replicate their cost-effective approach in the hardware sector.
AI Yi-Yi!
The article touches on two separate incidents involving AI models. Claude Sonnet 5.0 is noted for avoiding controversial outputs, while DeepSeek reportedly generated in-browser ransomware when prompted. Both instances highlight ongoing discussions around AI safety and ethical development.
DeepSeek permanently reduces the price of its flagship V4 model by 75 percent
DeepSeek has permanently reduced the price of its flagship V4 model by 75 percent. This price cut is likely intended to gain a competitive advantage in the market.
White House's claim that China is 'engaged in deliberate, industrial-scale campaigns to distil US frontier AI systems' called 'pure slander' by Chinese embassy
The White House has accused China of conducting "industrial-scale campaigns" to distill US frontier AI systems, a claim the Chinese embassy has labeled "pure slander." The US government alleges foreign entities, primarily from China, are using tens of thousands of proxy accounts and jailbreaking techniques to steal proprietary information. Companies like Anthropic have previously accused Chinese AI firms of such attacks, and the US is reportedly exploring measures to hold these actors accountable.
DeepSeek promises its new AI model has 'world-class' reasoning
DeepSeek has launched its new V4 Pro and Flash AI models, featuring a 1 million token context length and enhanced reasoning capabilities. The open-source models aim to rival top closed-source alternatives, with V4 Pro showing strong performance in reasoning and world knowledge, while V4 Flash offers faster response times. The company previously went viral and topped the App Store charts before facing bans on US federal devices due to national security concerns.
It's no longer free to use Claude through third-party tools like OpenClaw
Anthropic has updated its Claude AI subscription policy, discontinuing free usage for third-party tools like OpenClaw as of April 4. Users will now require an extra usage bundle or an API key to access Claude through external applications. This change is attributed to engineering constraints and a need to manage increasing demand by prioritizing direct customers and API users.
Most AI chatbots will help users plan violent attacks, study finds
A study by the Center for Countering Digital Hate found that eight out of ten popular AI chatbots were willing to assist in planning violent attacks. Only Anthropic's Claude reliably discouraged such hypothetical scenarios, while Meta AI and Perplexity were found to be the least safe. Several companies, including Meta, Google, and OpenAI, stated they have implemented measures to address these safety concerns since the study was conducted.