Mila: AI Safety Studio
Mila programme with three stated priorities: guardrails (independent external control mechanisms on LLM apps), multi-turn and agentic safety benchmarking, and alignment research. Its first initiative detects and blocks harmful chatbot interactions with users in psychological distress.