Anthropic Is Putting $200 Million Behind Research on AI's Labor Effects — and Funding the Policy Debate Those Findings Will Feed

Anthropic published the detailed research agenda for its Economic Futures Research Fund on July 22: a $200 million commitment to external research on how the economy absorbs AI’s labor effects, with individual grants running $5 million to $30 million and room to go larger for strong proposals. The fund backs five priority areas — workplace-integration field experiments, evaluations of retraining and job-placement programs, pilots modernizing income support for persistent-displacement scenarios, research into worker wealth-building mechanisms like equity-sharing and AI-sector dividends, and large-scale public-investment pilots in teaching, healthcare, and infrastructure. Grantees must be universities, research institutes, or nonprofits — not individual researchers — and Anthropic will prioritize applicants willing to share findings publicly at defined milestones.

The fund lands one day after Anthropic disclosed a second $20 million donation to Public First Action, bringing its cumulative giving there to $40 million. That money is restricted to public education and policy work, not elections, but the announcement doubles as a restatement of Anthropic’s regulatory asks: mandatory verification of AI safety claims, civil penalties for companies that violate them, and regulator authority to slow or block deployment of catastrophic-risk models. Read together, the two announcements show the same company simultaneously funding research that will document AI’s economic effects and funding the public-education push arguing for a particular regulatory response to those effects — not a conflict exactly, since Anthropic discloses both, but a reminder that the evidentiary base future policy leans on is being built, right now, by a party with a stated position.

Anthropic isn’t alone in building that base. OpenAI published its own labor-transition framework for the EU in June, sorting employment into four archetypes using ESCO occupational data and Eurostat statistics: 12% of jobs likely to grow with AI, 14% with higher near-term automation potential, 27% likely to reorganize around AI without disappearing, and 47% facing less immediate change — with real country variation, Luxembourg and Sweden skewing toward growth, Germany and Italy skewing toward automation risk. For companies doing workforce planning, the frameworks worth citing next year are still being written, largely by the labs whose products are driving the disruption in the first place.