Anthropic faces Musk’s ire after claiming its AI models were ripped off

In a sector where technology develops at the pace of a galloping cheetah, Anthropic put out an announcement on Monday saying some of its AI models were ripped off. How? In mass-scale attacks by AI companies that perhaps don’t grab headlines as much as OpenAI, Anthropic, xAI, and other leading firms. 

“We’ve identified industrial-scale distillation attacks on our models by DeepSeek, Moonshot AI, and MiniMax.These labs created over 24,000 fraudulent accounts and generated over 16 million exchanges with Claude, extracting its capabilities to train and improve their own models,” read the X post. 

The update comes as both OpenAI and Anthropic continue to butt heads, with both CEOs refusing to hold hands in a group photograph clicked at the India AI Summit, which had India’s Prime Minister Narendra Modi in attendance. 

The announcement said DeepSeek, Moonshot AI, and MiniMax used a method called distillation, which means to train newer, usually smaller-scale, AI models using existing AI models for purposes of imitation. 

While Anthropic says this is common practice in the industry if carried out within a firm for developing newer models, they take issue that it was done “illicitly”, which raises quite a few concerns for them and the U.S. economy. 

Some of these concerns are loss of key technology to China, newly-developed AI models without necessary safeguards, and an increased possibility of malicious cyber-web activity if the AI model is placed in the hands of criminals. 

More than a week ago, Anthropic declared a massive funding round of $30 billion, giving it a valuation of $380 billion. 

Due to its substantial backing by a huge group of investment firms, Anthropic is now one of the most valued AI startups in the world alongside OpenAI, primarily owing to its strong support from big players in America’s capital markets—a system which is capable of funding startups with billions of dollars in capital over a time-line. 

Notable institutional and corporate investors in the firm are: 

• Lighspeed Venture Partners 
• NVIDIA 
• Google 
• Amazon
• Microsoft  
• Fidelity Management & Research
• Iconiq Capital
• Menlo Ventures

Alongside significant backing from outside investors and companies, Anthropic’s CEO Dario Amodei and his sister Daniela Amodei hold significant equity in the startup, alongwith a handful of other founders and employees that include Jared Kaplan, Jack Clark, Chris Olah, Tom Brown, and Sam McCandlisk. 

Despite raising the alarm on X, Anthropic is getting no sympathy from the community, especially Elon Musk, who owns rival firm xAI, which is now absorbed within SpaceX. 

“Anthropic is guilty of stealing training data at a massive scale and has had to pay multi-billion dollar settlements for their theft. This is just a fact,” wrote Musk on X, after which he went on to repost multiple other tweets, in which users echoed the same sentiment of not wanting to give Anthropic any leeway due to accusations of the AI firm stealing huge amounts of data in the first place, rendering their own set of complaints moot. 

Anthropic said it would be defending against future attacks with the following measures: detection, intelligence sharing, access controls, and countermeasures. 

“As we noted above, distillation attacks at this scale require a coordinated response across the AI industry, cloud providers, and policymakers. We are publishing this to make the evidence available to everyone with a stake in the outcome,” wrote Anthropic.