Anthropic says that a group of Chinese AI startups are “distilling” their models by setting up huge numbers of fake accounts for Claude AI. In a blog post, Anthropic said that it disrupted “industrial-scale” campaigns by Chinese AI labs DeepSeek, Moonshot, and MiniMax. The company said that the group had over 16 million exchanges with Claude, after setting up 24,000 “fraudulent” accounts. Anthropic said it is developing countermeasures to prevent such attacks in the future.
Microsoft shares dropped as it announced a revised agreement with OpenAI.
The amended agreement ends revenue-sharing payments from Microsoft to OpenAI, and also ends Microsoft’s exclusive access to OpenAI’s intellectual property (i.e. models and products).
OpenAI’s revenue sharing with Microsoft will end in 2030, is subject to a total cap, and is no longer dependent on its achieving artificial general intelligence.
Amazon, a likely beneficiary of this lack of exclusivity, initially popped on the news but erased those gains.
This is a developing story.
China has ordered Meta to unwind its $2 billion acquisition of Manus, a Chinese startup (since relocated to Singapore) that makes AI agents and was central to Meta’s push to turn its massive AI investments into a real business. The move is part of the Chinese government’s effort to stop US firms from gaining access to Chinese talent and intellectual property, as Washington continues to restrict sales of advanced AI chips to Chinese companies.
Unlike its tech peers, which can sell AI through cloud services, Meta mainly uses AI to improve its existing ad business rather than as a stand-alone revenue driver. The decision strips away one of Meta’s clearest paths to monetizing AI — leaving it spending like a hyperscaler, without a hyperscaler business model.
Unlike its tech peers, which can sell AI through cloud services, Meta mainly uses AI to improve its existing ad business rather than as a stand-alone revenue driver. The decision strips away one of Meta’s clearest paths to monetizing AI — leaving it spending like a hyperscaler, without a hyperscaler business model.
Chinese AI lab DeepSeek has released a major new version of its eponymous open-source AI models that are nipping at the heels of leading frontier models in some areas.
The most significant DeepSeek-V4 Pro and DeepSeek-V4 Flash both have a 1 million-token context — the amount of information the model can actively work with in a single session — which is a crucial feature for complex, long-running coding tasks.
DeepSeek rebuilt how the models process information under the hood, making them substantially more efficient — and that efficiency is what makes the large context window actually usable.
Also, the new models’ coding skills have closed the gap with the major frontier models from Anthropic, OpenAI, and Google.
The authors of the model acknowledge some of V4’s shortcomings, such as its lower scores on reasoning benchmarks, saying that V4 “trails state-of-the-art frontier models by approximately 3 to 6 months.”
As open-weight models, V4 can be run on any user’s own hardware, making the V4 models among the top-performing open-source models out there. V4’s large context and token efficiency are especially significant among open-source models.
But like with earlier DeepSeek models, don’t ask it about Tiananmen Square.
DeepSeek rebuilt how the models process information under the hood, making them substantially more efficient — and that efficiency is what makes the large context window actually usable.
Also, the new models’ coding skills have closed the gap with the major frontier models from Anthropic, OpenAI, and Google.
The authors of the model acknowledge some of V4’s shortcomings, such as its lower scores on reasoning benchmarks, saying that V4 “trails state-of-the-art frontier models by approximately 3 to 6 months.”
As open-weight models, V4 can be run on any user’s own hardware, making the V4 models among the top-performing open-source models out there. V4’s large context and token efficiency are especially significant among open-source models.
But like with earlier DeepSeek models, don’t ask it about Tiananmen Square.
