American startup Anthropic has launched a broad market offensive, presenting the Claude Cowork toolset and the Remote Control mobile application. Simultaneously, the company officially accused Chinese competitors, including DeepSeek and MiniMax, of mass violation of service terms through systematic downloading of Claude model responses for training their own systems. Tensions are intensified by reports from US administration officials, according to which DeepSeek is training its latest systems on advanced Nvidia Blackwell processors, suggesting circumvention of export restrictions.

Anthropic's Corporate Offensive

The introduction of Claude Cowork and Remote Control aims to monopolize automation services for investment banking and engineering.

Accusations of Data Theft

Anthropic and OpenAI accuse China's DeepSeek of illegally using their models to train its own AI systems.

Nvidia Chips in China

The White House confirmed that DeepSeek uses banned Nvidia processors to build its next breakthrough model.

The artificial intelligence market has entered a phase of open technological and legal conflict. Anthropic, one of the sector leaders, presented Claude Cowork – a solution designed to automate complex processes in engineering, finance, and human resources. The launch of the Claude Opus 4.5 model and the Claude Code tool for autonomous programming garnered significant interest from developers, though it did not translate into sharp increases in the stock prices of software companies on the stock exchanges. Investors welcomed the expansion of the offering with specific business tools with relief, despite earlier concerns about AI's impact on the traditional IT service models of companies like IBM. Simultaneously, Anthropic leveled serious accusations against Chinese entities. According to the company, DeepSeek and MiniMax models were trained on data generated by the Claude bot, constituting a blatant violation of intellectual property. The situation is exacerbated by a White House statement confirming that DeepSeek has access to the latest Nvidia graphics chips. This discovery calls into question the effectiveness of US sanctions aimed at limiting China's access to strategically significant technology. Since 2022, the United States has been systematically tightening export limits on advanced chips to China, fearing their use for military and surveillance purposes. Beijing responds to these steps by intensifying domestic research and building alternative supply chains. In the face of growing competitive pressure, Anthropic is placing increasing emphasis on protecting its models from data theft, sparking lively debate among experts. The company declares continued commitment to ethics, but the intense race with OpenAI and Chinese rivals forces an ever-faster pace of innovation implementation while simultaneously tightening access to technology. Meanwhile, the CEO of Workday noted ironically that despite the rivalry, both Anthropic and OpenAI operate on his company's systems, which shows deep interconnections within the entire tech ecosystem. „We are seeing systematic harvesting of our models by Chinese firms, which undermines the foundations of fair competition in the AI sector.” — Dario Amodei

Mentioned People

  • Aneel Bhusri — CEO of Workday, commented on the dependence of AI startups on traditional software.