Why it matters
For teams building production AI, this shift towards specialised agentic models for security signifies a move away from general-purpose LLMs in sensitive domains. It highlights the importance of domain-specific fine-tuning for reliability in automated threat detection.
Key points
- Public launch of Project Perception scheduled for 3 August
- Introduction of MAI-Cyber, a model purpose-built for cybersecurity tasks
- Focus on agentic security to automate complex defensive workflows
- Industry focus on vendor-reported performance metrics for the new model



