What happened
According to a post trending on Hacker News, Anthropic is blocking the accounts of users who engage in any activity related to training larger models. The post is based on a short social media message, and no official company statement is included in the source.
Why it matters
When large language models are trained on another model's outputs, a practice known as "distillation" comes into play. Providers typically restrict this in their terms of service, since they do not want their own models used to train competing ones.
Open questions
- It is unclear what criteria the blocks are based on.
- There is no data on how many users are affected.
- Anthropic's confirmation of the matter is not present in the source.
For this reason, the story should be treated as a claim until an official announcement clarifies it.
