Anthropic Accuses Chinese AI Rivals of Industrial-Scale Theft of Claude Technology in Explosive New Claims

Anthropic Accuses Chinese AI Rivals of Industrial-Scale Theft of Claude Technology in Explosive New Claims

As governments worldwide grapple with AI export controls and chip restrictions, Anthropic’s revelations add fresh urgency to the conversation about protecting frontier AI technology in an era of rapid, borderless innovation.

RMN News Service

New Delhi, February 24, 2026 — AI pioneer Anthropic has leveled serious accusations against three leading Chinese artificial intelligence laboratories, claiming they orchestrated “industrial-scale distillation attacks” to steal advanced capabilities from its flagship Claude model.

The U.S.-based company, creator of the popular Claude chatbot, says DeepSeek, Moonshot AI, and MiniMax set up more than 24,000 fraudulent accounts to generate over 16 million interactions with Claude. The goal, according to Anthropic, was to “distill” — or extract and repurpose — its most sophisticated features, including agentic reasoning, tool use, and advanced coding, into their own competing models.

In a strongly worded statement, Anthropic warned that such illicit distillation bypasses built-in safety guardrails and poses significant national security risks. The company highlighted concerns that the stolen capabilities could be integrated into foreign military, intelligence, and mass-surveillance systems, potentially enabling authoritarian regimes to accelerate offensive cyber operations and disinformation campaigns.

Also Read:

[ Sam Altman “Confused” as India’s AI Ambitions Face “Identity Crisis” and Leadership Scrutiny ]

[ From Novel to Transmedia IP: Robojit’s AI-Assisted Production Pipeline ]

[ India’s AI Identity Crisis: Mirage of Progress ]

“These campaigns are growing in intensity and sophistication,” Anthropic said. “The window to act is narrow.”

The allegations come as Anthropic celebrates a spectacular financial milestone: a massive $30 billion Series G funding round that has catapulted the company’s post-money valuation to an eye-watering $380 billion. The fresh capital follows the recent launch of Opus 4.6, Anthropic’s most powerful model yet, designed to automate complex professional workflows including document generation, spreadsheet analysis, and full presentations.

The timing of the accusations has sparked immediate backlash — most notably from Elon Musk, CEO of rival xAI. Taking to his platform X, Musk fired back at Anthropic, accusing the company of hypocrisy. He claimed Anthropic itself is “guilty of stealing training data at massive scale” and has paid multi-billion-dollar settlements related to the creation of Claude. Netizens quickly piled on, with many calling the situation “deeply ironic” given that major AI firms, including Anthropic, have trained their models on vast swaths of publicly available (and often copyrighted) internet data.

Despite the heated public debate and questions over data practices across the industry, Anthropic’s momentum appears undiminished. The company’s explosive growth underscores the fierce global race for AI supremacy, where accusations of intellectual-property theft are becoming an increasingly common flashpoint between Western and Chinese labs.

As governments worldwide grapple with AI export controls and chip restrictions, Anthropic’s revelations add fresh urgency to the conversation about protecting frontier AI technology in an era of rapid, borderless innovation.