Is it actually risky or against the rules to put client tax returns into AI tools, as some critics on social media claim?

The group discussed a recurring pattern on platforms like X where some accountants publicly criticize others for putting tax returns into AI tools, implying legal or professional risk (e.g., "you're going to go to jail") without explaining what the actual problem is or how to do it correctly. The consensus among speakers was that nobody actually knows definitively what the risks are or how this should be handled - not even AI company leadership, since even Anthropic's own CEO reportedly doesn't have a clear read on how AI will play out. Speakers pointed out apparent double standards: Adobe already shows an AI tool banner when you open sensitive documents, Google has been scanning emails for ad targeting for years, and clients already routinely email their own Social Security numbers to their accountants - all of which are widely accepted. The group's stated approach going forward is to push back on unsubstantiated criticism by asking critics to explain specifically what they know and how it should be done, rather than accepting vague warnings at face value.