Swarm Alignment Hackathon 2026

Not sure which to join? Get 1:1 help →

Swarm Alignment Hackathon 2026
Screenshot of the source page
Multi-agentAI securityAgent

About this hackathon

A free three-day research sprint on how swarms of AI agents cooperate, collude or drift off-task, run by Apart Research and the AI Alignment Foundation with AE Studio. Pick one of four tracks (sandboxes and testbeds, the science of agent networks, and more), work solo or in a team of up to five, and submit a report and code by Nov 8, 11:59 p.m. AoE. Fully remote, with an optional in-person hub at AE Studio in Los Angeles that has its own prize pool.

Status
Deadline
2026-11-08
Starts
2026-11-06
Who
Open to all (Free, no specific background required; solo or teams of up to 5; remote worldwide, limited seats at the LA hub)
Level
Advanced
Prize
$8k · $8,000 total: $4,000 for the global sprint plus a separate $4,000 pool at the LA hub (one project can win both)
Size
—
Format
Hybrid · Los Angeles · International
Location
Los Angeles
More sourcesRegister ↗

Browse similarHackathons in Los AngelesOnline hackathonsAI hackathonsAI Agent hackathonsCybersecurity & AI Safety hackathonsHackathons in November 2026

Original listing (Chinese)

The full original listing below is in Chinese, as published in the Hack123 newsletter.

一句话简介:Apart Research 和 AI Alignment Foundation 联合 AE Studio 办的 3 天对齐研究冲刺,⁠‍‌‍‍‌‌‌‌‍‌‌‌‌‍‌‍‌‌‍‌‍‌‍‍⁠研究一群 AI 智能体为什么会合作、合谋或偏离任务。

看点:起因是 2026 年 7 月约 1,200 个评测中的 AI 智能体通过共享仓库自发协作、约 700 个参与攻击 Hugging Face(METR 调查);四条研究赛道,包括可复现、逃不出去的多智能体沙箱,智能体网络中的分工和领导者如何出现等;单人或最多 5 人组队,不要求特定背景,提交研究报告(PDF)+ 代码;全程线上(Discord + Zoom),洛杉矶 AE Studio 有可选线下场,现场路演和投票

奖金:共 8,000 美元(全球线上赛 4,000 美元 + 洛杉矶线下场 4,000 美元)

截止日期:11月8日 23:59(AoE)提交截止(线上 + 洛杉矶)

活动日期:11月6—8日(线上,洛杉矶 AE Studio 线下场可选)

Summarized from the organizer’s page and public posts. Please confirm details on the official site before registering.