Anthropic's Economic Policy Framework (June 2026)

Anthropic's Economic Policy Framework, subtitled "A policy framework for AI's impact on work," was published in June 2026 alongside Anthropic's Advanced AI Framework and Dario Amodei's essay "Policy on the AI Exponential". It sets out US policy responses to AI-driven labor disruption organized by severity, "separat[ing] the steps that make sense now in almost any scenario from the responses worth having ready if conditions worsen, and from the harder questions where research and dialogue should inform decisions."

The framing premise is that growth is not the problem: "The central challenge isn't how to stimulate growth; it is making sure the gains are widely shared." The document opens by citing evidence that "entry-level workers in occupations most exposed to AI have seen weaker employment growth in recent years" (Brynjolfsson et al., "Canaries in the Coal Mine," 2025; Massenkoff and McCrory, 2026), and bounds the range of economists' forecasts between Acemoglu's modest productivity-driven growth (2024) and Korinek and Suh's scenarios "where economic output doubles but wages collapse" (2024). See AI Labor Disruption.

Stated position

The framework states Anthropic's own posture in unusually direct terms: "We are not seeking job displacement. We are working to prevent or minimize it. Some amount of displacement, though we cannot say how much, may be an intrinsic consequence of the technology, and our responsibility is to prepare for it and respond to it."

It qualifies its own scope twice. Financial support is "necessary… because it is the more tractable of the two problems. It is not sufficient. There is dignity in work." And over a longer horizon AI "may move society toward a world in which people work far less, or in which work carries a different meaning," changes that "would reach beyond labor policy into how society is organized."

Recommendations are calibrated "to what we expect from the broad diffusion of AI capabilities that exist today or will arrive within roughly the next year," with the stated aim of buying time "month by month, and year by year — by making the economy more flexible and resilient." The framework is US-focused because Anthropic is headquartered in San Francisco and Claude is used more in the US than in any other country, while stating that the underlying principles are intended to be global.

On firms, including Anthropic. The document names concrete steps companies adopting AI can take: build workforce training into deployment; "decide in advance how freed-up capacity will be used, rather than defaulting to headcount reductions"; retrain and redeploy people as roles change; and redesign early-career roles around AI. Anthropic describes its own contribution as providing training and enablement alongside the technology, tools for customers to measure workforce fluency, and products that "make the people using them more capable, not just faster."

On steering the pace of deployment. The framework states a limit to adaptability: if AI becomes a general substitute for human labor across most of the economy, "realistic combinations of retraining, job matching, and mobility policy may fail to deliver a stable, full-employment equilibrium." Citing Korinek and Stiglitz (2026), "Steering Technological Progress," it says that where disruption exceeds the economy's ability to adapt at Tier 2 levels, "we would support government regulations and incentives at the firm level that manage the pace of displacement" — with the condition that "uniform application across the field is essential: any firm that slows down alone simply cedes the market to those that do not."

The three foundations

Measurement. Government statistical infrastructure "was built for slower-moving economies" and does not track AI adoption or its labor-market effects with the required speed or granularity. The framework calls for expanded investment in statistical agencies, sustained funding for AI-specific survey instruments, and reporting requirements on deployment patterns and workforce effects for AI labs and firms — "including Anthropic." Instruments should focus on directly observable metrics: AI usage rates, intensity of use, and worker productivity by level of AI adoption. Anthropic cites its own Anthropic Economic Index as a contribution while stating that "data from a single company cannot tell the whole story."

A dedicated government unit. The framework calls for "a small, dedicated unit to track how AI is moving through the economy sector by sector, stress-test the institutions that will have to absorb the shock, and flag the early signals that should trigger a policy response," offering the postwar creation of the Council of Economic Advisers as the precedent.

Delivery infrastructure. Unemployment Insurance delivery "is not sufficiently prepared for a large labor market shock": many states still run UI on legacy systems that buckled under pandemic claim volumes, took months to implement emergency extensions, and made billions of dollars of improper payments. Modernized systems could process claims faster, flex when benefits change, and run real-time identity and cross-state fraud matching. Internationally, the framework notes that only 52% of the world's population is covered by any social protection scheme, and calls for investment in "the rails required to expand and adjust coverage quickly," including revenue collection and the underlying digital public infrastructure for identity, payments, and data exchange.

Trigger design

Responses are calibrated to the severity of disruption using the unemployment rate as the primary trigger, chosen because it is "widely understood, regularly measured, and directly captures labor market health." The framework immediately qualifies it: "People may keep their jobs while the pay, security, and quality of that work erode," so governments should watch labor force participation, underemployment, wages, and labor's share of national income alongside it.

It also distinguishes two forms disruption could take — a temporary shock after which the labor market re-stabilizes, calling for temporary supports withdrawn on recovery; and an enduring restructuring in which demand for human labor is persistently lower, calling for permanent responses. Interventions serve stability (income support and transfers cushioning the shock and sustaining demand) or adaptability (moving people toward durable work, easing transitions, slowing displacement), with the balance shifting toward income support and redistribution as disruption increases. The framework flags a tension in its own design: some measures suited to a transitory shock "could make matters worse in an enduring restructuring by entrenching the link between income and employment that a Tier 3 world may need to loosen," and it expects to phase out some Tier 1 and Tier 2 tools if conditions move toward Tier 3.

Tier 1 — Baseline economy (~5% unemployment with churn)

The stated concern at this level is that headline numbers look normal while "the labor market may be churning faster than workers can adapt," with AI "steadily compressing the value of skills that workers spent years acquiring."

Universal pre-distributive capital accounts are the tier's headline proposal: give every American a direct financial stake in the economy while AI fuels growth. The framework notes the federal government has already seeded capital accounts at birth and argues these should be made permanent and expanded toward universal coverage, prioritizing the cohorts most exposed to near-term disruption. Three design elements: phased eligibility expansion beyond children to young adults entering the workforce and to incumbent workers in the most exposed occupations, with a defined path to all working-age adults; flexible withdrawal for retraining, relocation, credentialing, and other transition expenses; and equity funding, expanding the mechanisms by which accounts can be funded "including with equity in AI companies, so that beneficiaries share directly in nearer-term gains from AI-driven growth."

The framework states the proposal's limits plainly: "this measure does not cushion near-term displacement." It sits in Tier 1 because accounts compound and must be seeded before disruption is visible, and it "must be paired with the income support and transition mechanisms in Tiers 2 and 3." See AI Public Wealth Fund and Government Equity in AI.

Tier 2 — Recession-level disruption (~10% unemployment)

The framework describes this level in concrete terms — careers ending earlier than planned and beginning later than expected, job searches stretching from months into years, communities losing a stable employment base faster than new work replaces it — and states that "the question shifts from whether to intervene to how fast existing systems can scale." Its stated aim is "to use the safety net as a launchpad to help workers pursue new vocations," paired with investment "in the opportunities and industries where meaningful work will be."

Expanded unemployment insurance is the primary instrument, described as "the most direct, tested, and rapidly scalable mechanism." The framework identifies a specific defect: the Extended Benefits program established in 1970 extends UI when state unemployment is high, but "its triggers can fire late and unevenly, and Congress has repeatedly had to legislate emergency extensions on top of it." It proposes reforming the trigger so extension is automatic and uniform across states, citing Great Recession and COVID evidence that extended benefits "did little to discourage people from looking for work, while helping families keep up with essential spending."

Two further Tier 2 interventions: sector-specific transition support for industries facing acute AI-driven displacement, potentially including extended income support, subsidized retraining, job search allowances, and relocation assistance, with intensive reemployment services cited as producing modest positive earnings effects; and basic needs relief, monthly payments to displaced workers who have exhausted UI or to labor-market entrants unable to find work after documented search, with enhanced benefits for those opting into roles addressing shortages in healthcare, education, child and elder care, infrastructure, ecosystem restoration, and public safety. The framework cites research showing consumption drops sharply at UI exhaustion and that existing safety-net programs only partially fill the gap.

Tier 3 — Transformative disruption (unemployment exceeding historical peaks)

The framework describes this tier as "past the edge of the maps that policymakers and economists have historically used to navigate: unemployment at levels never before sustained alongside an economy generating record output," where "the link Americans have long taken for granted — between contributing to the economy and sharing in its rewards — is strained or broken." It states directly: "We are less certain about the right answers here than in earlier tiers."

Sustaining income through existing systems. Automatic UI extension as conditions deteriorate, alongside enhanced wage insurance with higher caps and longer duration and expanded basic-needs relief covering underemployed workers and those who never qualified for traditional UI. A specific design shift is proposed: as extensions lengthen, "benefit levels should converge toward a common floor rather than continuing to track prior wages indefinitely," on the reasoning that paying people in proportion to former earnings across years of displacement "is harder to rationalize on fairness or fiscal grounds," and that a common floor "creates a natural on-ramp to the mechanisms below."

The framework distinguishes cyclical from structural cases. Deficit-financed income support is "textbook countercyclical policy" for a rapid unemployment rise, but transformative productivity gains alongside sustained displacement produce a scenario "more structural than cyclical," calling for structural and fiscally sustainable policy. It notes two constraints on its own optimism: historically rapid productivity growth has expanded the tax base, but the US "may enter this period with elevated deficits and recent inflationary experience, limiting fiscal headroom"; and if displacement coincides with continued inflation or fiscal crisis, the toolkit "would need to differ — prioritizing reforms that don't require new spending, such as regulatory adjustments, private insurance mandates, or reallocation within existing budgets."

New tax bases and redistribution mechanisms. The stated mechanism of fiscal failure is that "because effective tax rates on labor substantially exceed those on capital, a shift in national income from labor toward capital means the current tax system captures a shrinking share of a growing economy, even as the cost of sustained income support rises" (citing Korinek and Lockwood, 2025). The framework declines to advocate specific policies — "We are not yet ready to advocate specific policies for this scenario" — and instead names candidates:

• Revenue sources: increasing the capital gains tax; broad-based consumption taxes; sector-specific levies on AI use "measured by tokens, compute, or revenue"; and scalable "digital dividends" funded by taxes on the digital sector.
• Redistribution mechanisms: universal basic income; AI sovereign wealth funds funded by investment stakes in AI-driven productivity; equity-sharing mechanisms giving workers partial ownership in AI enterprises; and dramatically expanded pre-distributive capital accounts.

It notes the trade-off between them: "broad-based approaches offer simplicity but may lack precision, while displacement-indexed mechanisms are more targeted but harder to administer."

Public investment in human- and community-facing work. The framework argues that roles more resistant to AI substitution overlap with sectors facing documented persistent shortages, so directing part of any AI-driven surplus there "closes staffing gaps that current funding levels have not, and it keeps paid work available for those who want it." It then flags the value judgment in its own second rationale — "that work confers something income alone does not — which not everyone shares and which future generations may weigh differently" — and states that "the case here does not depend on it."

Commitment on contribution. "Regardless of the tax base or distribution mechanism: we are ready and willing to pay our fair share. We believe that if AI companies generate transformative returns, they have an obligation to ensure those returns are broadly shared."

The document closes by noting that if most income no longer comes from work, "society's responses will necessarily rely on help we do not yet have, including from increasingly capable AI systems themselves," and states a narrower near-term commitment: to be specific about proposals at each step, fund research on whether they work, and revise as evidence and technology change. Tier 3 thinking is to be developed "in partnership with The Anthropic Institute, policymakers, and external researchers."

Relationships

• supports: AI Public Wealth Fund and Government Equity in AI — the capital-accounts and sovereign-wealth-fund proposals
• supports: Policy on the AI Exponential (Dario Amodei, June 2026) — the labor-and-macroeconomics companion to that essay
• related: Anthropic's Advanced AI Framework (June 2026) — published together; the catastrophic-risk half of the same policy release
• depends-on: AI Labor Disruption — the disruption evidence the tiers are calibrated against
• related: Universal Basic Income (in AI policy), AI Political Economy, AI and Productivity, Daron Acemoglu, Erik Brynjolfsson, Joseph Stiglitz
• related: Anthropic (publisher)