Is security/compliance a concern for the Open Accountant project, and what about using custom/open-source AI models instead of Claude?

The response was that security in the sense the question was framed isn't quite the right concern for Open Accountant, because the project isn't about using Claude to compute taxes -- it's built for people who are already using Claude to compute taxes correctly. On the topic of alternative models, it was mentioned that in a prior session another member described using their own custom models in their firm, which reportedly benefited the firm significantly. Hugging Face was recommended as a resource -- described as a collection of open-source models -- and the team said they are currently testing a few models from there themselves. The suggestion for firms interested in this route was to look into Hugging Face and consider appointing someone internally to help set up and maintain a custom model.