Entropy Alignment Security Hackathon Berlin 2026, Sponsored by Anthropic
Registration closed · Similar hackathons still open
- Swarm Alignment Hackathon 2026$8k · Due 11-08
- Naïve Autonomous Company HackathonDue 11-07
- BAND Dark Factory Buildathon: San Francisco & Remote 2026$4.5k · Due 11-13
- Drexathon: Nace AI Decision Model Hackathon San Francisco 2026Due 11-14
- NEAR HACKATHON 2026$250k · Due 11-14
- SF AI Research Hackathon (Iterate x Boardy)Due 11-14
Not sure which to join? Get 1:1 help →
About this hackathon
A one-day, in-person AI security and alignment hackathon in Berlin on Nov 14, 2026 (10:00–22:00 CET, w3.hub Kreuzberg), sponsored by Anthropic, Hack The Box, Circuit and Nebius. About 100 people in 20 teams of five attack a purpose-built vulnerable AI range to find, exploit and fix LLM weaknesses, across three tracks: securing recursive autonomous LLM loops, detecting covert channels and collusion in multi-agent systems, and an open track for novel prompt injections and guardrail bypasses. Every participant gets $100 in Anthropic credits. Prizes are $20,000 in Claude credits ($10K / $6K / $4K). Apply on Luma; the hosts pick a balanced room.
- Status
- Deadline
- 2026-11-14
- Starts
- 2026-11-14
- Who
- Open to all (For AI security and alignment researchers, ML engineers and CTF players; apply on Luma and the hosts select about 100 people; teams of five, solo applicants get matched)
- Level
- Advanced
- Prize
- $20,000 in Claude credits: $10K 1st, $6K 2nd, $4K 3rd; every participant also gets $100 in Anthropic credits
- Size
- ~100 people
- Format
- In person · Berlin · International
- Location
- Berlin
Browse similarHackathons in BerlinAI 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：Entropy 在柏林办的一日 AI 安全与对齐黑客松，Anthropic、Hack The Box、Circuit、Nebius 赞助，在专门搭好的漏洞靶场里找出、利用并修补大模型漏洞。
看点：约 100 人、20 支 5 人队，单人报名可现场匹配；三条赛道：保护自主递归执行的 LLM 循环、检测多智能体之间的隐蔽信道与串谋、开放赛道（新型提示词注入和护栏绕过）；按新颖性、影响和对企业 AI 风险的现实意义评分；每人送 Anthropic 100 美元额度，全天管饭；Luma 申请，主办方按背景挑人
奖金：2 万美元 Claude 额度，第一名 1 万、第二名 6,000、第三名 4,000
截止日期：11月14日当天开赛（Luma 申请审核制）（柏林，线下）
活动日期：2026年11月14日 10:00—22:00（柏林时间，北京时间 17:00—次日 05:00；柏林 w3.hub）
Summarized from the organizer’s page and public posts. Please confirm details on the official site before registering.