Cohere Labs, Lexsi Labs and Hugging Face Launch AI Safety Hackathon

Key Info

Lexsi Labs, Cohere Labs and Hugging Face are partnering on the Lexsi AI Safety and Alignment Hackathon, running online and free from Nov 1 to 21, 2026, open to builders worldwide.

Highlights

• Teams build real post-training pipelines using open-weights Tiny Aya and Aya Expanse models from Cohere Labs
• Tooling includes six open-source Lexsi Labs kits: AlignTune, CuratorKIT, AuditKIT, SafeTune, CircuitKIT and AgentTune
• Artifacts such as models, adapters, logs and datasets are hosted on Hugging Face
• Tracks cover fine-tuning and data curation, AI red teaming and safety