Lexsi AI Safety and Alignment Hackathon

Registration closed · Similar hackathons still open

All open hackathons →

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

Lexsi AI Safety and Alignment Hackathon
Screenshot of the source page
AI securityAgentData

About this hackathon

A free, fully online hackathon by Lexsi Labs with Cohere Labs and Hugging Face: chain Lexsi's open-source post-training, red-teaming, safety and interpretability toolkits to make a small open-weights Cohere model measurably better or safer. Teams of 1–3; register by Nov 5, submit by Nov 14, Demo Day Nov 22. Prizes are $30K+ in Lexsi/Cohere credits, HF Pro and HF GPU credits, plus interviews.

Status
Deadline
2026-11-05
Starts
2026-11-01
Who
Open to all (Open worldwide, from undergrads and beginners to researchers; teams of 1–3, one team per person)
Level
Intermediate
Prize
$30,000+ in prizes and credits (Lexsi/Cohere credits, HF Pro and HF GPU credits; no cash): 1st place $1,750 each in Lexsi and Cohere credits plus 6 months HF Pro per teammate; $200 HF credits for up to 20 finalists; $20 HF credits for up to 800 teams; interview opportunities
Size
—
Format
Online · International
Location
Bengaluru (online; placed at organizer’s city)

Browse similarOnline hackathonsAI hackathonsAI Agent hackathonsData 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:Lexsi Labs 联合 Cohere Labs、Hugging Face 办的全球线上黑客松:用 Lexsi 开源的后训练、红队、安全修复与可解释性工具链,⁠‌‌‌‌‌‌‍‍‌‍‍‍‍‍‌‌‍‌‍‌‍‍‍‌⁠把 Cohere 的小型开源权重模型调得更好、更安全,并用前后指标证明。

看点:三条赛道(微调与数据整理、AI 红队与安全对齐、智能体微调);须串联同赛道至少 2 个开源仓库,模型、数据与日志托管在 Hugging Face;1—3 人组队、免费参赛;入围 Demo Day 的最多 20 队;表现突出者有 Lexsi Labs 实习或研究岗面试机会

奖金:总计 $30,000+ 奖品与额度(Lexsi / Cohere 额度、HF Pro 与 GPU 额度,无现金);第一名每项 $1,750 额度+全队 6 个月 HF Pro

截止日期:11月5日报名截止(AoE);11月14日作品提交截止(线上,全球)

活动日期:11月1日—14日线上开发;11月22日 Demo Day(班加罗尔线下+线上)

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