AGI × Animals Wargame

By Sentient Futures (formerly AI for Animals), Constance Li @ 2025-10-08T02:14 (+62)

In September 2025 in Warsaw prior to CARE, 19 animal movement leaders came together to participate in the first ever AGI and animal welfare wargame. 

TLDR: Great-power moves and food-security shocks repeatedly overrode animal welfare efforts; standard campaigning tactics like mass mobilization struggled under political crackdowns and were overshadowed by government’s concerns about public unrest in the face of war risks.

Executive Summary

What this was

A strategic tabletop exercise designed to test whether current animal-advocacy theories of change held under fast AI progression and geopolitical stress. Participants were selected based on their potential strategic influence on the direction of the animal movement and assigned roles according to their pre-existing domain expertise which ranged from governments, frontier AI labs, animal agriculture companies, alternative-protein companies, animal advocates, funders, and media. They were then given an [amateurly] forecasted starting scenario and asked to stay in character while interacting and negotiating with one another over 4 rounds and 3 hours. 

This exercise was adapted from the AI-2027 tabletop exercise, which we ran once exactly “as is” to understand the game mechanics. We then ran it again with a newly written animal welfare scenario for Compassionate Futures Summit before making additional modifications to make the pre-CARE game as useful and smooth as possible. 

Why we ran it

Sentient Futures is a field-building organization focused on steering the future toward the welfare of all sentient beings—most of whom, as far as we know, are nonhuman animals. Because of the scale and urgency of animal suffering, interventions often default to triage, and scenario planning is rare in this space. Other cause areas already use it (e.g., ALLFED’s nuclear-winter food resilience work; The Artificial Intelligence Underwriting Company’s commercial approach to addressing AI risk). We ran this wargame to bring that discipline into animal advocacy and to test whether current strategies would remain robust and scalable as AI advances. 

The outcomes summarized below are intended to surface likely constraints, plausible leverage points, and common failure modes for animal advocacy and food-systems change under AGI dynamics.

Contents of this write up

This write up covers the major events that happened during the game, reflections on how existing theories of change held up (or didn’t), and shares the materials and instructions for replicating the game if you want to run it for your own group.

The Game

Scenario in brief

It was January 2029. The U.S. had a Vance administration. China was rolling out the Golden Wing Initiative, an AI-enabled precision livestock farming (PLF) system for caged broilers, domestically and across parts of Africa. Frontier labs kept accelerating. After the 2nd round, it was announced that U.S. labs had successfully developed AGI.

See here for the full starting scenario.

Game Highlights

• China’s frame landed: China was praised for delivering cheap, stable protein at scale to Africa. When the EU blocked Golden Wing chicken on regulatory grounds, China offered a cage-free variant to unlock access. The “development + biosecurity + affordability” narrative worked well.
• U.S. hard line, tech-first posture: The U.S. banned cultivated meat by executive order in the first round and prioritized AI dominance. It took an anti-China stance in refusing Golden Wing imports, which prompted U.S. animal-ag companies to build their own PLF systems.
• Food security trumped welfare: Alternative proteins gained ground in most places except the U.S., where the early ban froze adoption. Wherever war risk, unemployment, or price spikes rose, food-security narratives overrode welfare and justice narratives.
• Alt protein drew broad investment and accelerated with AI. We saw multiple governments and private capital commit to alt-protein R&D and manufacturing. Escalating war risk and food-security concerns made many countries treat it as strategic infrastructure, while a U.S. executive ban stalled domestic development. As AI capabilities advanced, product development, process optimization, and supply-chain scaling moved markedly faster.
• Windfall philanthropy from labs: Animal-sympathetic employees at frontier AI labs became extremely wealthy and channeled hundreds of millions into the movement thanks to early relationships with animal funders. Some advocacy groups effectively received blank checks as a result.
• Early opposition in Africa under-performed: Animal advocates launched massive anti–Golden Wing campaigns across several African regions. The consumer effect was minimal against price and availability. An attempted undercover investigation in China was unsuccessful.
• Coalitions mattered when funding broke. Welfare and rights groups formed an early coalition that became essential after major funders paused support to rights organizations and some groups saw their 501(c)(3) status revoked or frozen amid government concerns about their activities' impacts on the U.S. economy, which was largely prompted by animal ag lobbyists.
• Funders hit a wall... and then a cliff: Frustration grew as progress stalled and the U.S. took away nonprofit status for some entities including foundations. One donor went rogue and attempted to assassinate the U.S. president. Media and segments of the public framed the attacker as an anti-hero (Luigi-Mangione-style coverage).
• Tech worker leverage vanished at AGI: Rights advocates organized a strike by vegan employees at a major AI company over use of its tech in animal agriculture; it worked—briefly. Once AGI capability was declared, those employees became replaceable and were ordered by the U.S. government to be fired.
• Welfare metrics inside PLF stalled: Welfare groups and animal-ag companies began defining key welfare indicators to monitor inside PLF systems. Without policy hooks, enforcement, or retailer commitments, that work lost momentum as energy flowed back into oppositional and reactionary campaigns.
• Narrow-AI vs AGI confusion: Several teams tried to share and redeploy AI tools across domains (e.g., alt-protein modeling to power advocacy targeting).
• Chicken pivot resistance in China: A government proposed pivot toward chicken (for greater feed-conversion efficiency to help with food security in times of war) ran into cultural resistance tied to pork. An open question emerged: would advocates pivoting towards reinforcing status quo pork preference be effective in slowing the small-body replacement problem?
• AI race between US-China escalated towards war: The U.S. tightened chip export controls. China accelerated domestic compute production and signed a deal with the EU for chip components. A short-lived pause agreement on AI advancement collapsed. China started preparing for war by stockpiling food reserves. By the final turn, bloc commitments and mobilization signals heavily hinted at WWIII.

Takeaways

Participant reflections

We asked participants for candid reflections on what felt realistic, what broke, and what could be improved. Key themes and selected paraphrased notes included:

• “The game made me sad.”
• “It was striking that both iterations of the AGI × Animals wargame ended the same way, with U.S. vs. China escalation and animal welfare sidelined.”
• “We were woefully underprepared as the animal orgs. I’m not sure anyone had done proper strategy prep, which would have been essential. These kinds of games could be useful for testing strategies once we have them.”
• “Normal social-movement strategies for animals lost all potency and didn’t have time to play out. It updated my view that we have a rapidly closing window of opportunity to influence the future for animals.”
• “More cooperation with animal-ag-tech actors around suffering reduction (genetics, welfare metrics/monitoring) seemed possible, but it also felt deeply unsatisfying to rely on those alone.”
• “The movement shouldn’t focus on what worked in the past ten years, but on what will work in the next five. Old theories of change aren’t built for the world we’re entering.”
• “Rogue actors and actions had a large footprint—but their effects were unpredictable.”

Observed themes

1. Convergence toward major-power conflict.
   Across all runs, endgames converged on U.S. vs. China escalation. By the final turns, we were approaching major-power war (“WWIII” trajectories).
2. Power flipped the board quickly.
   Executive actions, export controls, and bloc deals reordered priorities rapidly. Prior regulations and animal-welfare commitments were easily reversed with little accountability from decision-makers. The early U.S. ban on cultivated meat had a devastating lock-in effect and removed a powerful lever for food systems change.
3. Food security dominated under stress.
   As war risk and economic instability grew, price stability and calorie reliability outweighed public appeals about animal welfare.
4. Windows were short.
   Opportunities appeared and closed quickly. Speed and adaptability mattered more than long, linear campaigns.
5. “Outrageous” events felt less implausible.
   Elements that seemed extreme or unrealistic during the game (e.g., a high-profile assassination; nonprofit status risk) were actually echoed by real-world developments. The day after the game, political activist Charlie Kirk was assassinated. Earlier in the year, an executive-order threatened removal of 501(c)(3) statuses.

Implications for the movement

• Conventional social-movement tactics like public pressure lacked time to mature. Coalitions helped preserve capacity when funding or legitimacy faltered.
• Future theories of change should aim to “ride the wave” of AI by being ready to mobilize workers displaced by AGI, planning for sudden post-AGI philanthropy windfalls, and aligning with narratives like food security which will become more important as geopolitical stress rises.
• Building relationships with AI companies proved essential for sustaining funding and legitimacy during economic instability. Making enemies with powerful actors could be existentially risky to all of our work. The potential loss or freezing of nonprofit status would devastate the current animal-advocacy ecosystem, which remains largely defenseless against such moves.

Limitations

A note on scope and use.

The outcomes and takeaways for this game are mostly meant as a jumping-off point for exploring potential strategies and is not meant to be a prediction of the future. It was only as good as the participants’ ability to model their characters’ perspectives. Real-world dynamics were hard to replicate, and many aspects of the gameplay were simplified to keep the game moving. There are things we would do differently if we ran the game again such as forcing more of the player actions to be focused on animals rather than on AI race dynamics. 

Appendix

Reference materials

Feel free to use these for running your own AGIxAnimals Wargame!

• Application Form - to recruit participants
• To Print:
  • Player ID Cards - to be printed on 13x19" poster printer, cut and inserted into these badge holders and put on lanyards for players
  • Probability Cards - printed via this site and one set distributed to each player
  • Starting Scenario and Player Cheat Sheet - one copy given to each role (pair of players playing each role)
• Intro slides - presented at start of game and contain gameplay instructions
• Participant Prep Packet - sent via email before game
• In Game Notes - displayed during game to track time and actions/outcomes
• Feedback form - sent after the game
• Recommended reading - should be emailed beforehand: Jamie Harris’s post outlining potential strategies for animal advocates in the face of transformative AI

For any questions or assistance with adapting materials, please contact game [at] sentient futures [dot] ai

Stay in the loop about Sentient Futures' programs including our fellowships, conferences, and community. 

Acknowledgements

Thank you to the many individuals who helped develop, iterate, and run this game: Ronak Mehta, Eli Lifland, Jonas Vollmer, Sanu Basurag, Jay Luong, Sam Chapman, Elena Voronkova, and Marta Mikita. 

We are especially grateful to The AI Futures Project for early guidance on game development and Anima International for generously covering the cost of hosting the event.