Copying accusations in the AI world

Anthropic has formally accused DeepSeek and other Chinese companies active in the field of artificial intelligence of copying its models on a large scale. The American company claims that the "distillation" process, i.e. the creation of smaller models from larger models, took place through the use of fraudulent methods.

Details of the accusations

According to Anthropic, the alleged copying would include the use of as many as 24,000 accounts created ad hoc and 16 million data exchanges. These would have been used to train smaller models, unduly benefiting from Anthropic's work and investments. At the moment, DeepSeek has not released statements regarding the accusations.

For those evaluating on-premise deployments, there are trade-offs that AI-RADAR analyzes in detail on /llm-onpremise.