Anthropic is testing a completely new steering option for large language models. The AI startup is offering developers access to its Beta Steering API, which can be used to customize the internal functions of language models. The API is based on recent research on the interpretability of language models. By strengthening individual concepts in models, their output can be strongly influenced.
Interested developers will get access to a subset of Claude's internal features, documentation, sample code, and possibly a Slack channel to communicate with the Anthropic team. In return, testers will be asked to share their projects with Anthropic and provide feedback. The technology is still in the research phase and is not intended for production use. Anthropic emphasizes that the API may be modified or discontinued at any time.
