**Defensive agents, blueprints & adversarial testing gaps**
Key Questions
What is Project Glasswing?
Project Glasswing is a coalition including Apple, Google, Anthropic, AWS, Microsoft, NVIDIA, and Cisco to secure AI systems from threats, offering $100M in credits for Mythos vulnerability hunting.
How is Trent AI addressing agentic security?
Trent AI raised $13M for the first multi-agent security platform, focusing on securing the agentic age with tools like Arctic Wolf Aurora.
What defensive efforts involve PANW and others?
Palo Alto Networks focuses on securing AI itself, alongside Acronis MDR, SonicWall NSM, and SANS CRISP for managed detection and response.
How does DOE contribute to AI hardening?
DOE's CESER shifts cyber R&D toward AI hardening via budget proposals, emphasizing governance and skills from RSAC.
What is the Security Claw framework?
Security Claw is an open-source, self-hosted AI agent framework designed for agentic AI evolution and defensive security operations.
What adoption rates exist for AI in security?
73% AI adoption rate with 88% shadow AI concerns, highlighting needs for VITG-SecCentric MDR and TrojAI platform extensions.
How do humans factor into AI cybersecurity?
RSAC 2026 emphasizes people in cybersecurity, with AI-led models like Arctic Wolf keeping humans in the loop for success.
What governance challenges arise with defensive AI agents?
Boards must manage AI-reshaping cyber risks, with Microsoft and RSA updates focusing on identity security in the AI age.
Glasswing coalition (Apple/Google/Anthropic/AWS/MS/NVIDIA/Cisco) Mythos vuln hunting/$100M credits; Arctic Wolf Aurora/PANW/Acronis/SonicWall NSM/SANS CRISP/Trent AI $13M multi-agent sec; Security Claw/TrojAI/OWASP; DOE CESER AI hardening; RSAC skills/governance; 73% adoption/88% shadows; VITG-SecCentric MDR.