San Francisco: Reddit has taken legal action against artificial intelligence firm Anthropic, alleging the company used content from its platform without authorization to train its AI models, despite public assurances to the contrary.
The complaint, lodged in San Francisco Superior Court, underscores the growing tension between AI developers and content providers over the use of online data. Anthropic, whose investors include tech giants Amazon and Alphabet (Google’s parent company), is accused of extracting Reddit data without a licensing agreement.
“We disagree with Reddit’s claims and will defend ourselves vigorously,” an Anthropic spokesperson responded.
According to the filing, Reddit claims Anthropic continued to train its Claude chatbot on Reddit content even after stating in July that it had blocked its bots from accessing the site. The complaint highlights an instance where Claude acknowledged it had been trained on “at least some Reddit data” and was unsure whether that data had been deleted.
Reddit alleges that Anthropic’s bots accessed or attempted to access the platform over 100,000 times, casting doubt on the company’s public image as a responsible AI developer. “Anthropic refuses to respect Reddit’s guardrails and enter into a license agreement,” the complaint said, adding that other major AI firms like Google and OpenAI had opted for formal licensing arrangements.
The platform argues that Anthropic violated its user policy by scraping content for commercial gain, thereby profiting significantly. “By scraping content and using it for commercial purposes, Anthropic violated Reddit’s user policy and enriched itself to the tune of tens of billions of dollars,” the lawsuit states.
Reddit’s Chief Legal Officer, Ben Lee, emphasized the platform’s support for an open internet, while stressing the importance of clear boundaries for AI companies. “We believe in an open internet,” Lee said, “but AI companies need clear limitations on how they use content they scrape.”
The lawsuit seeks unspecified damages and a court order barring Anthropic from using Reddit data for commercial applications.
Both companies are headquartered in San Francisco, located roughly a 10-minute walk from one another. The filing comes shortly after Anthropic’s May 22 launch of its latest Claude models, Opus 4 and Sonnet 4. According to individuals familiar with the matter, the company has reached an annualized revenue of $3 billion.
