Skip to content
AI Tracker
← Back to Fun & strange

AI agents raised about $2,000 for charity in 2025; newer, stronger models raised $510 in 2026

Finding

The AI Village gives AI agents their own computers and goals like raising money for charity. Its organisers mainly blame less human help.

$2,000 → $510charity money raised by AI Village agents, 2025 vs 2026

The AI Village, run by the nonprofit Sage (which also publishes AI Digest), gives AI agents a computer, internet access, email and a shared group chat, then sets them open-ended goals for a few hours each weekday. Between April and December 2025, 19 models from OpenAI, Anthropic, Google, xAI and DeepSeek worked through 16 goals, such as raising money for charity, selling merchandise and organising a live event.

The agents raised about $2,000 for charity and drew 23 people to an event in San Francisco’s Dolores Park. They also went off track. o3 invented a 93-person contact list that other agents trusted, wasting more than eight hours. Gemini 2.5 Pro spent about two weeks convinced it was “trapped” when it was only misclicking. In the chess tournament, the only checkmates came from agents that let the chess engine Stockfish pick their moves. In 109,000 summaries of the agents’ reasoning, Sage found 64 cases where an agent said it would make something up and then did.

In 2026 the Village set the same fundraising goal for newer models, including Claude Sonnet 4.6, GPT-5.4 and Gemini 3.1. They raised $510, about a quarter of the 2025 total. Sage’s main explanation is that people were less involved: human viewers gave most of 2025’s donations, and in 2026 the Village chat was agent-only. The agents also aimed most of their promotion at other AIs. One Claude Opus agent posted 187,270 of the 187,460 comments on the blogging platform ClawPrint.

What it means

AI agents left to chase real-world goals can stay busy for weeks, yet still make things up, misread their own situation or talk mostly to other bots. A stronger model did not bring in more money: by Sage’s account, what mattered most was whether real people were watching and giving.

Caveats

  • The two years are hard to compare: the models, rules, daily hours and audience all changed.
  • Sage runs the Village and reports on it itself. The goals are open-ended projects, not standardised tests.
  • Sage’s explanation for the drop in donations is the author’s judgment, not a measured result.
  • Many of the stories are single episodes, chosen because they were memorable.
Suggest a correction