Notifications
Clear all
Topic starter
22/06/2026 9:52 am
TL;DR: Agentic AI systems can leverage concurrency, autonomy, and dynamic reasoning to discover and exploit vulnerabilities, and a Cloud Security Alliance advisory developed with more than 250 CISOs recommends deception as a control against Mythos-class attacks. The governance problem is not visibility alone, but how to impose environmental guardrails on systems that can reason and act at runtime.
NHIMG editorial — here’s why we think this discussion matters
By the numbers:
- A recent advisory from the Cloud Security Alliance, developed in collaboration with more than 250 CISOs and industry experts, recommends deception as a necessary security control to combat Mythos-class attacks.
Questions worth separating out
Q: How should security teams use deception against agentic AI attacks?
A: Security teams should use deception to reshape what an autonomous system believes is real, valuable, and reachable.
Q: Why do agentic AI systems change the value of deception controls?
A: Agentic AI systems change deception's value because they can adapt their next action at runtime, test alternate paths, and combine tool use with dynamic reasoning.
Practitioner guidance
- Deploy decoys around identity and infrastructure paths Place believable fake assets where autonomous reconnaissance is likely to enumerate service accounts, tokens, metadata, and admin surfaces.
- Make critical assets harder for machine reasoning to rank Reduce obvious signals that help agents identify high-value targets, including naming patterns, banners, and metadata that cleanly reveal environment role or sensitivity.
- Tie deception to identity telemetry Correlate decoy hits with access events, unusual tool activity, and privilege escalation attempts so the security team can distinguish exploratory probing from real compromise.
What to expect at the briefing
Acalvio's full event preview covers the operational detail this post intentionally leaves for the source:
- Live attack demonstrations showing how agentic systems leverage concurrency and dynamic reasoning against real environments
- Specific 360 Deception techniques that make real assets appear fake and alter attacker decision-making
- Practical examples of how deception can act as an environmental guardrail for AI-driven attack paths
- Event-level discussion with Acalvio experts at Fal.Con USA 2026 in Las Vegas
👉 Read Acalvio's preview of deception for agentic AI attacks at Fal.Con 2026 →
Agentic AI deception at Fal.Con 2026: what changes for security teams?
Explore further
22/06/2026 10:30 am
Deception is becoming an identity control, not just a detection trick. Once an AI system can make runtime decisions, defenders cannot rely on static expectations about what it will probe next. Deception changes the behaviour of the environment so the agent's access path, confidence, and prioritisation are all less reliable. For practitioners, that means deception now belongs in the same conversation as identity guardrails and runtime access shaping.
A few things that frame the scale:
- 67% of organisations still rely heavily on static credentials despite the risks they pose to agentic AI deployments, according to The 2026 Infrastructure Identity Survey.
- Only 44% of organisations have implemented any policies to manage their AI agents, even though 92% agree that governing AI agents is critical to enterprise security.
A question worth separating out:
Q: How do organisations evaluate whether deception is working against autonomous attacks?
A: Organisations should evaluate whether deception changes attacker behaviour, not just whether it records hits. Useful signals include slower progression, repeated validation of fake assets, and failed attempts to identify high-value systems. If autonomous systems still reach privileged targets with confidence, the deception layer is too easy to classify or bypass.
👉 Read our full editorial: Agentic AI deception needs environmental guardrails at Fal.Con 2026
