AI Giants Face Lawsuits Over Stolen Content

Prominent publishers have initiated legal action against AI giants OpenAI and Microsoft, asserting that hundreds of thousands of their articles were systematically stolen to train AI models like ChatGPT and Copilot. The lawsuit, filed in the U.S. District Court for the Southern District of New York, highlights the substantial market value generated by these AI products, claiming no revenue has been shared with the content owners whose work enabled their creation. AI companies, conversely, maintain that the internet-wide scraping of data falls under existing 'fair use' provisions.

This legal battle is not isolated. Anthropic has also leveled accusations, sending a letter to federal officials alleging that Chinese tech firm Alibaba illicitly used its Claude AI model. Between late April and early June, Alibaba reportedly utilized nearly 25,000 fraudulent Claude accounts for exchanges that served as training data for its own AI system through a process known as adversarial distillation.

Anthropic has previously voiced similar concerns regarding other Chinese AI startups, including DeepSeek, Moonshot, and MiniMax, and OpenAI has also accused DeepSeek. While these actions may not be definitively illegal, Anthropic argues they violate terms of service and necessitate a coordinated response from the U.S. public and private sectors to maintain a competitive edge in AI development.

Microsoft is also exploring alternatives, with its AI division head announcing a new model, MAI-Thinking-1, trained without any distillation. This approach aims to avoid potential biases inherited from 'teacher' models. The ongoing disputes, similar to publisher lawsuits, suggest a prolonged and complex resolution process for intellectual property and AI development.