Vai al contenuto
AI.info

events

Entropy Alignment Security Hackathon

Un hackathon di un giorno a Berlino in cui 20 squadre sfruttano e correggono vulnerabilità degli LLM nei filoni dedicati all’allineamento, agli agenti e alla prompt injection.

Entropy Alignment Security Hackathon | Eventi sull’IA

In inglese

Starts
2026-11-14 09:00 UTC
Ends
2026-11-14 21:00 UTC
Organizer
Personal
Location
Berlin, Germany

Entropy hosts a one-day hackathon in Berlin that brings together 100 AI security and alignment specialists to find, exploit and address vulnerabilities in large language models. Twenty teams of five work on a vulnerable AI range built for the event, with tracks on securing autonomous LLM loops, detecting covert channels and collusion between agents, and novel prompt injections and guardrail subversion. Anthropic, HackTheBox, Circuit and Nebius sponsor; prizes are Claude credits. Solo participants can be matched into teams. It is aimed at alignment security researchers, ML engineers and CTF players.

Event website