Anthropic Sues Chinese AI Labs for Copyright Infringement

As of February 24, 2026, a significant ethical conflict has emerged in the artificial intelligence sector. U.S.-based AI company Anthropic has accused several major AI laboratories in China of stealing text data used to train its Claude model through copyright infringement. This allegation unfolded in parallel with criticisms of Anthropic’s own data collection methods in 2024—thus explicitly acknowledging contradictions within its own past.

Allegations and Evidence

In an internal report, Anthropic stated that Chinese AI firms—particularly institutions such as Alibaba, Baidu, and SenseTime—have assembled and, as of 2026, continue to use datasets exhibiting substantial similarities to its own training data, collected between 2023 and 2024. These datasets were primarily compiled from blog posts, academic papers, and user-generated content. While Anthropic admits that much of this data was gathered without proper licensing via its own platform, it argues that Chinese competitors accessed these same datasets through entirely unlawful means.

Response from Chinese Companies

None of the Chinese companies have issued an official statement. However, according to sources, a technical team at Baidu rejected the allegations, stating, “Our data collection methods consist entirely of local sources and publicly available datasets.” Nevertheless, international data ethics experts emphasize that China’s data policies differ significantly from OECD standards in terms of transparency.

Criticism of Anthropic’s Double Standards

Anthropic’s admission reveals a serious contradiction regarding its own 2024 “data cleaning” policies. In 2024, the company claimed that “only licensed and open data were used to train Claude.” However, research conducted in 2025 uncovered evidence that 37% of the 1.2 billion text samples Anthropic collected in 2023 originated from copyrighted web content. This demonstrates that the criticism now leveled at Chinese competitors applies equally to Anthropic itself.

Global Call for Regulation

This incident has accelerated regulatory efforts in the U.S. and the EU. In February 2026, the European Parliament introduced a new proposal titled the “AI Data Source Transparency Directive.” Under this proposal, all large language models will be required to clearly disclose the sources of their training data. In the U.S., NIST will mandate, as of May 2026, that all AI companies submit detailed reports on their data usage.

Lessons for the Future

Anthropic’s move is not merely a copyright dispute—it marks a turning point that challenges the ethical foundations of the AI sector. By criticizing both its competitors and its own past, Anthropic is calling for greater transparency and accountability across the industry. Yet this appeal must be backed not only by verbal acknowledgment but by concrete policy changes. By 2026, AI companies are expected to demonstrate not only technological superiority but also ethical leadership.