Taxing AI to Help Workers Sounds Good, but Public Deserves More 向AI征税以帮助工人听起来不错，但公众应获得更多

TL;DR

• Rep. Greg Casar's proposed AI Tax and Work Protection Act taxes AI token usage to fund worker support, but the article argues this is a flawed proxy for taxing automation itself
• The token-based tax creates practical enforcement challenges, including valuing an unstable technical unit and attributing unemployment spikes to AI versus other causes
• An equity-based approach, where governments take ownership stakes in AI companies, is proposed as a superior alternative that captures economic gains without continuous valuation of AI usage
• The article references Professors Jeremy Bearer-Friend and Sarah Polcz's proposal for in-kind equity taxation and Sen. Bernie Sanders' more aggressive sovereign wealth fund legislation
• Equity ownership scales with actual economic gains from AI, requiring only a one-time company valuation rather than ongoing assessment of token usage and labor displacement causation

Why It Matters

This analysis directly addresses a critical policy question at the intersection of AI governance and economic distribution: how should societies capture and redistribute the economic gains from AI-driven automation? For AI practitioners and policymakers, the distinction between taxing AI use versus taxing AI ownership has profound implications for how the industry will be regulated and how its economic benefits will be distributed. The equity-based framework could reshape the competitive landscape for major AI companies and establish new precedents for technology taxation globally.

Technical Details

• Casar's bill imposes a levy equal to the greater of: (1) a percentage of fair market value of tokens processed in covered transactions, or (2) a percentage of revenue and related-party value associated with those transactions, with the tax rate keyed to the unemployment rate
• The proposed "safety valve" allows the Treasury Department to adjust the tax escalator when unemployment traces to war, pandemic, or unrelated economic shocks — a causal allocation mechanism the article identifies as practically unworkable
• Bearer-Friend and Polcz's alternative proposes in-kind taxation where generative AI companies transfer equity directly to the government rather than paying cash taxes
• Sanders' version would require covered AI companies to transfer a 50% stake to a federal sovereign wealth fund, though the article suggests a smaller, independently managed equity assessment limited to the largest firms
• The article notes that tokens are not standardized commodities sold in arm's-length markets, making them an unreliable unit for tax valuation, whereas company equity requires only periodic valuation rather than continuous per-transaction measurement

Industry Insight

• AI companies should anticipate equity-based taxation frameworks as a realistic policy direction and begin structuring ownership and governance to accommodate potential government stakeholder participation
• The token-tax approach, while politically appealing for its feedback loop mechanism, creates regulatory uncertainty around valuation methodology that could burden companies with compliance complexity and unpredictable tax liabilities
• The equity-based model creates a natural alignment between public benefit and AI company performance, potentially encouraging more responsible deployment while giving companies an incentive to demonstrate broad economic value beyond labor displacement
• Policymakers should focus on defining clear coverage thresholds (which firms qualify), valuation methodologies, and governance structures for public equity stakes rather than attempting to meter and tax individual AI computations

TL;DR

• Casar的AI税收法案以token为征税对象存在估值难题和因果归因缺陷
• 股权分配方案能让公众分享AI经济收益，避免持续评估token价值
• 主权财富基金模式（如Sanders提案）是更可行的政策替代路径
• 政策设计需平衡政府监管能力与企业合规成本

为什么值得看

该分析揭示了AI政策设计中征税机制与自动化实际影响之间的错配问题，为从业者理解监管趋势提供了关键视角。股权分配方案可能重塑AI产业的经济收益分配格局，影响企业战略与公众接受度。

技术解析

Casar法案采用双轨征税机制：按AI token交易的市场价值或相关收入较高者计征，税率与失业率挂钩形成反馈循环。但token作为技术单位缺乏标准化定价，且需实时归因失业与AI的因果关系，执行难度高。

股权方案主张企业以实物股权替代现金缴税，政府一次性获得公司份额后持续分享收益。该设计避免了对每次AI使用的估值，转而依赖企业整体价值评估，更符合自动化经济影响的长期性特征。

Sanders提案将股权分配扩展至主权财富基金模式，要求覆盖企业转让50%股权。政策建议采取更温和路径：仅针对头部企业征收较小股权份额，通过独立投资工具管理，降低政府监管负担。

行业启示

政策制定者应优先设计基于经济收益而非技术使用量的AI影响评估框架，避免陷入持续估值与因果归因的操作困境。AI企业需提前规划股权分配等新型合规路径，将公众收益分享纳入长期战略。监管创新可能推动AI产业从"使用税"向"收益共享"模式转型，重塑行业竞争格局。