US-based Anthropic has implemented new defenses to prevent competitors from reverse-engineering its Claude AI, explicitly accusing Chinese firm Zhipu AI of unauthorized data harvesting. The move highlights deepening geopolitical tensions in the technology sector. Investors tracking global AI trends should note Zhipu AI’s significant stock price correction in Hong Kong and its ongoing financial challenges, including widening losses and heavy research spending.
US-based artificial intelligence lab Anthropic has deployed new security layers to stop competitors from practicing 'model distillation.' This technique involves using the outputs of a highly advanced system, like Anthropic's Claude, to train smaller, cheaper models, effectively reverse-engineering the original technology. Anthropic reported that these new safeguards, which block access to its reasoning tokens and proxy networks, have reduced unauthorized data extraction by up to 99 percent.
In a rare public move, Anthropic identified Chinese startup Zhipu AI as a primary entity behind these attempts. The US company alleges that Zhipu AI used automated accounts to scrape Claude’s capabilities to develop its GLM-5.3 model. This public accusation underscores the intense competition and geopolitical friction surrounding artificial intelligence, where technological supremacy is often viewed as a matter of both economic and national security.
For investors following the global tech landscape, the situation highlights the significant financial and operational pressures facing AI developers. Zhipu AI, which is listed on the Hong Kong Stock Exchange (ticker: 2513) and not available on Indian exchanges like the NSE or BSE, has faced considerable volatility. Its share price has corrected significantly from a peak of approximately HK$2,980 in June 2026 to around HK$643 in early October 2026.
The company’s latest financial data reflects the high cost of maintaining AI infrastructure. For the first half of 2026, Zhipu AI reported revenue of RMB 953.89 million but faced a net loss of RMB 2.07 billion. This gap highlights the difficulty many AI firms face in balancing rapid revenue growth with the immense capital spending required for research and development, particularly while operating under strict export controls that limit access to high-end global chips.
The primary risks for investors in this sector remain tied to regulatory scrutiny and operational constraints. Because Chinese AI firms often rely on domestic compute clusters due to US-led chip export restrictions, their ability to match the efficiency and reasoning power of Western counterparts may be hampered. Furthermore, as large AI labs like Anthropic take aggressive legal and technical stances against data scraping, the costs of maintaining competitive models may rise, putting additional pressure on profit margins.
Moving forward, investors may track how these defensive measures by US companies influence the operational costs and development timelines for AI firms in China. Key monitorables include future regulatory actions, the ability of companies like Zhipu AI to manage their high cash burn, and whether such technical barriers permanently limit the ability of competitors to replicate proprietary reasoning architectures.