AI Agent Security Could Become One of the Most Valuable New Cybersecurity Skills

Amazon Research Awards' Fall 2026 call for proposals, which opened on 1 October, includes AI for Information Security and Amazon Security among its research areas. Together they highlight where important AI-security problems are emerging.

Research topics across these calls include:

AI-agent access governance
Authentication and authorisation
Agent guardrails
Intrusion and anomaly detection
AI-powered incident response
Agent red-teaming
Vulnerability detection
Secure AI toolchains
Secure code generation

These are academic research calls, not a measure of the whole job market. But they provide a useful career signal.

As organisations deploy more autonomous AI systems, cybersecurity professionals increasingly need to understand how agents authenticate, what resources they can access, which actions they can perform and how those actions can be monitored.

A valuable technical combination is emerging:

Cybersecurity + AI + Agentic Systems

Amazon's submission period runs through 4 November 2026.

Takeaway: Securing AI agents is becoming a real technical discipline involving traditional security principles plus new agent-specific risks.