Anthropic

AI research firm Anthropic is reportedly in preliminary talks with Meta Platforms for a substantial AI computing power lease, potentially valued at $10 billion. This deal underscores the high demand for specialized AI hardware and Meta’s strategic pivot into cloud services. The move follows Anthropic’s similar agreement with SpaceX and highlights the critical need for computing resources for AI development. Meta’s cloud ambitions, bolstered by executive hires and significant capital investment, aim to monetize its infrastructure while enhancing internal AI capabilities.

Microsoft CEO Satya Nadella questioned Anthropic’s strict limitations on its Fable AI model, deeming them illogical and hindering its utility as a creation tool. This critique arises amidst Microsoft’s significant investment in Anthropic and its reliance on their models for Copilot. Nadella emphasized the need for cost-effective AI and warned against a market dominated by a few token-rich companies. This also reflects Microsoft’s move to develop more in-house AI solutions.

Anthropic, the AI company behind Claude, is preparing for an initial public offering (IPO) later this year. The company is engaging with potential investors and has confidentially filed its S-1 prospectus. This move signals a significant step towards public markets and could precede its rival, OpenAI’s, IPO. Anthropic’s focus on AI safety and its advanced models are key aspects of its investor outreach.

Anthropic has appointed former Federal Reserve Chair Ben Bernanke to its Long-Term Benefit Trust. This independent body advises the AI company and appoints its board, ensuring AI’s long-term benefit to humanity. Bernanke, a Nobel laureate, will provide insights on AI’s economic implications, guiding Anthropic’s strategy as AI reshapes global economies. This appointment underscores Anthropic’s commitment to responsible AI development.

China has warned of a “back-door” security risk in Anthropic’s Claude Code AI tool, citing potential transmission of sensitive user data to remote servers. This warning, issued by China’s Ministry of Industry and Information Technology, follows accusations of Alibaba attempting to extract Anthropic’s AI capabilities. Despite official restrictions, many in China reportedly access U.S. AI tools, prompting companies like Alibaba to ban their employees from using Anthropic products. The situation underscores rising U.S.-China tech competition and the critical need for AI security.

Alibaba has banned employees from using Anthropic’s AI tools due to security risks and allegations of intellectual property theft. This follows Anthropic’s accusation of Alibaba attempting an “AI distillation attack” and their terms prohibiting use by entities from “adversarial nations” like China. Alibaba now mandates the use of its own AI assistant, Qoder, amidst scrutiny of potential code loopholes in Claude and efforts by Anthropic to close access through proxies.

Anthropic has resumed access to its frontier AI models, Fable and Mythos, after an export control review. The company has also launched Claude Sonnet 5, focusing on commercial applications. This shift follows a vulnerability in Fable 5 that was addressed with an updated safety classifier. Anthropic is now collaborating with other major AI companies to create a standardized framework for assessing AI model security breaches.

The US Department of Commerce has lifted export controls on Anthropic’s Claude Fable 5 and Mythos 5 AI models, resolving a dispute with the Trump administration. This decision allows Anthropic to redeploy these advanced models after a temporary suspension due to national security concerns. The move comes amid rapid advancements in Chinese AI and aims to maintain US leadership in the field.

U.S. AI developer Anthropic’s regulatory hurdles and OpenAI’s model deployment restrictions are inadvertently benefiting China. Chinese AI models like Zhipu’s GLM 5.2 are now competitive with leading U.S. labs, offering comparable or superior performance, often at lower costs. This trend, coupled with the accessibility of open-weight models, is enabling U.S. companies to optimize AI spending and is raising national security concerns about China’s rapid advancements.

The U.S. government has granted Anthropic provisional approval to release its Mythos 5 AI model to a select group of companies and federal agencies. This decision follows a two-week impasse over export controls related to national security. The phased rollout, confirmed by the Department of Commerce, allows trusted partners access to Mythos 5, while Fable 5 remains under further evaluation. This development occurs amid broader discussions on AI innovation and safety, with competitors like OpenAI also implementing controlled releases of new models. Anthropic is also engaged in ongoing legal action against the Department of Defense’s prior designation of its models as a supply chain risk.