Anthropic's Economic Policy Framework: the three-tier plan for AI and jobs, explained

The same month it launched Claude Fable 5, its most capable model, Anthropic published a document of a different kind: an Economic Policy Framework, dated June 2026, that sets out in writing the policy response to a scenario where AI disrupts the labour market for the long term. An AI lab spelling out what should happen if its own technology displaces jobs at scale is unusual, and worth reading closely, especially if you run a business.

This article breaks down what the document contains, the three tiers, the headline measures and Anthropic's own commitments, then what it means for companies deciding how to deploy AI now. Quotes come from the full framework (PDF, 14 pages). This version was checked against it on 17 October 2026.

The starting point: signals that are already measurable

The framework starts from research, not science fiction. It cites evidence that, in the US, entry-level workers in the occupations most exposed to AI have seen weaker employment growth in recent years, and notes that economists' models range from modest productivity-driven growth to scenarios where economic output doubles but wages collapse. Its core sentence: if capabilities keep advancing, "AI may become a general substitute for human labor", and the central challenge is not how to stimulate growth but how to make sure the gains are widely shared.

The document also states values: "We are not seeking job displacement", "There is dignity in work", and the most-quoted commitment, "we are ready and willing to pay our fair share", since "if AI companies generate transformative returns, they have an obligation to ensure those returns are broadly shared".

Three foundations before any measure

• Measurement. Statistical agencies do not track AI adoption or its labour-market effects with the speed or granularity needed. Anthropic calls for AI-specific survey instruments and reporting requirements for AI labs and firms, "including Anthropic", on deployment patterns and workforce effects. It contributes its own data through the Anthropic Economic Index.
• Analysis. A dedicated government function to track how AI moves through the economy and flag the signals that should trigger a response; the document points out that the Council of Economic Advisers was created for the post-war transition, and that an equivalent is needed for AI.
• Delivery. Many US states still run unemployment insurance on legacy systems that buckled under pandemic claim volumes, took months to implement emergency extensions and made billions of dollars of improper payments. Globally, the document notes, only 52% of the world's population is covered by any social protection scheme.

The core of the plan: three tiers indexed to unemployment

The framework's originality is its structure: a graduated escalation calibrated to the unemployment rate, watched alongside labour force participation, underemployment, wages and labour's share of national income. With each tier, the weight shifts from adaptability toward income support and redistribution.

Tier	Trigger, in the document	Logic	Headline measures
Tier 1	Baseline economy, about 5% unemployment with churn	Prepare for disruption, share AI's upside	Universal pre-distributive capital accounts, wage insurance, occupational licensing reform, retention tax incentives
Tier 2	Recession-level disruption, about 10% unemployment	Expand temporary income support and retraining	Automatic, uniform unemployment insurance extensions, sector-specific transition support, basic needs relief
Tier 3	Transformative disruption, unemployment exceeding historical peaks	New mechanisms for redistribution	Benefits converging toward a common floor, new tax bases, candidate mechanisms such as universal basic income and AI sovereign wealth funds

Tier 1: give everyone a capital stake before the storm

The headline measure is counter-intuitive: universal pre-distributive capital accounts. The US federal government has already created capital accounts seeded at birth; Anthropic wants them made permanent and expanded towards universal coverage, with phased eligibility starting with the cohorts most exposed to near-term disruption. Alongside: wage insurance, a time-limited, capped supplement for workers who take lower-paying jobs after displacement, occupational licensing reform and retention tax incentives.

The logic of Tier 1 is to act while the headline numbers still look normal: at 5% unemployment, the document says, the labour market may be churning faster than workers can adapt.

Tier 2: the safety net as a ramp back to work

At 10% unemployment, the response pivots on unemployment insurance. Today's federal extension "can fire late and unevenly", so the document proposes reforming the trigger so that extension is automatic and uniform across states, and argues that extended benefits did little to discourage job search during the Great Recession and the COVID pandemic.

It adds basic needs relief for those who remain without work after a documented search, ensuring access to food, housing and healthcare, with enhanced benefit levels for those who opt into roles addressing labour shortages in healthcare, education, child and elder care, infrastructure, ecosystem restoration and public safety.

Tier 3: when the link between work and income breaks

The last tier describes a world "past the edge of the maps that policymakers and economists have historically used to navigate": unemployment never before sustained, alongside an economy generating record output. Here Anthropic names candidates rather than recommendations, and says plainly that it is "not yet ready to advocate specific policies for this scenario":

• New tax bases. Because "effective tax rates on labor substantially exceed those on capital", a shift of national income toward capital leaves the current tax system with a shrinking share of a growing economy. Candidates: higher capital gains tax, broad-based consumption taxes, sector-specific levies on AI use, measured by tokens, compute or revenue, and "digital dividends" funded by taxes on the digital sector.
• Redistribution mechanisms. Universal basic income, AI sovereign wealth funds funded by stakes in AI-driven productivity, equity-sharing mechanisms giving workers partial ownership of AI enterprises, and much larger capital accounts.
• Public investment in human- and community-facing work, where the US already faces documented, persistent shortages.

The money on the table

Alongside the framework, Anthropic announced two investments totalling 350 million dollars: a 200 million dollar Economic Futures Research Fund, to fund research trials and evaluations of promising public policies, and a 150 million dollar national fellowship programme for early-career people extending the benefits of AI to communities across America (Anthropic).

The passage every leadership team should read

The framework addresses policymakers, but one passage speaks directly to companies. AI can be used to produce the same output with fewer people, or to accomplish more with the same people, by building new products, serving markets that were out of reach and raising the capability of less-experienced workers. The document lists four concrete steps for firms:

1. Build workforce training into deployment.
2. Decide in advance how freed-up capacity will be used, "rather than defaulting to headcount reductions".
3. Retrain and redeploy people as roles change.
4. Redesign early-career roles around AI, the roles the cited research shows are most exposed.

It goes further: if disruption went beyond the economy's ability to adapt, at the levels described in Tier 2, Anthropic says it "would support government regulations and incentives at the firm level that manage the pace of displacement", applied uniformly, because "any firm that slows down alone simply cedes the market to those that do not".

Our reading: three takeaways for a growing business

What follows is our analysis, not the document's content.

1. The signal matters as much as the plan. When the lab selling one of the most capable models on the market writes that AI "may become a general substitute for human labor" and puts 350 million dollars behind the research, that is information about the trajectory its leadership considers plausible. A business planning three years ahead should price it in, the way it prices in an interest-rate or energy scenario.

2. The adaptation window is now. The framework's logic is to "buy time" while the economy is still in Tier 1. For a company, the parallel is direct: the moment to train a team, redesign roles and decide what freed-up capacity is for comes while AI is still an advantage, not yet a condition of survival.

3. The four steps are an audit checklist. Training built into deployment, a plan for freed-up capacity, retraining, early-career roles redesigned: an honest grid for judging any AI project, yours or the one a vendor is pitching. A project that answers none of the four is probably a headcount plan dressed up as a transformation. That is the grid we apply to the AI features in our own products.