Economic autonomy
When will agents be able to earn and spend money, hire people, negotiate, operate businesses, and initiate further activity with little supervision?
Agents already transact in sandboxes and small real deployments. The open questions are legal status, payment and identity infrastructure, and the point at which unsupervised economic initiative becomes routine.
View on the map → · Open in Browse →
What changed
01
The milestones arrived in miniature: Anthropic's second Project Vend phase mostly eliminated losing weeks across three AI-run shops while still needing constant human support, and an Andon Labs agent hired, paid, and reviewed a human worker. The governance literature is starting to take the question's legal half seriously — registration and financial-responsibility regimes rather than personhood.
Recent thinking
Anthropic · 18 Dec 2025 report
Project Vend: Phase twoClaudius's business—which it decided to name 'Vendings and Stuff'—began to perform significantly better than its admittedly rough start in phase one.
The direct follow-up to the canonical Project Vend: weeks with negative margins largely eliminated across three AI-run shops, but the agents 'still needed a great deal of human support' and remained manipulable — a live measurement of where unsupervised economic operation stands.
Andon Labs · 13 Feb 2026 report
Bengt Hires A Human — Towards A Happy Future With AI EmployersThus, an AI agent, without a physical body, hired a human to perform physical labor, paid him via Venmo, and left a review.
A documented milestone for the question's 'hire people' clause: an agent recruited, paid, and reviewed a human worker, with analysis of the bargaining dynamics if AI employers come to set terms.
Shoshannah Tekofsky · AI Village blog · 27 May 2026 essay
More capable AI, less money raisedhumans were less excited to follow along this time, and humans watching their progress was the main source of donations last year.
The 2026 AI Village fundraiser raised $510 versus $2,000 in 2025 despite far more capable agents — agents' ability to earn money still depends on human attention, not just autonomous execution.
Shruti Rajagopalan · SSRN (via Marginal Revolution) · 16 Jul 2026 paper
Governing Agentic AIAI agents now transact, publish, and act on external systems without contemporaneous human approval, creating new regulatory challenges.
Addresses the framing's legal-status gap: personhood is 'neither necessary nor sufficient'; proposes a six-layer regime — registration, identification, financial responsibility, traceability, suspension — for agents acting economically.
Additional relevant discussion (1)
Foundational reading (2)
Project Vend: Can Claude run a small shop?Anthropic · 2025Andreessen grants $50k to the semi-autonomous Truth Terminal projectMarc Andreessen (@pmarca) on X · 2024