CPA Trendlines Podcasts Before Building with AI, Firms Need a Plan for Client Data | Innovation Insights
Oct 29, 2026
Liz Mason, CPA, is a tax technology entrepreneur and CEO of Taxfyle, High Rock Accounting, and Rebel Rock. She explores how firms can build AI tax tools without compromising privacy, accuracy, or professional ethics. Topics include IRS compliance, cyber liability, privacy vaults, model selection, human review, safe automation, environmental impact, and value-based pricing.
AI Snips
Chapters
Transcript
Episode notes
Put Compliance Before Your AI Pilot
- Start with IRS Regulation 7216, human review, security, and professional ethics before running an AI pilot.
- Liz Mason built proprietary OCR because general LLMs struggle to read tax documents visually and reliably.
AI Creates Risks Deterministic Software Does Not
- AI differs from deterministic SaaS because it makes statistical judgments and may not reproduce the same behavior consistently.
- A secure enterprise subscription does not automatically eliminate storage, terms-of-service, or unexpected-use risks.
Treat AI Uploads Like Emailing Tax Returns
- Treat uploading client data to general-purpose AI as a cyber-risk decision, not as equivalent to using an encrypted tax portal.
- Review breach consequences, notification obligations, cyber insurance, and deductibles before accepting that risk.