[CLSTR]
< Back to all clusters
[TECHNOLOGY] · United Arab Emirates, United States · 19 sources

started · updated

Part of situation: Security risks and emerging governance in agentic AI (28 clusters)

AI Agent Security: Rapid Deployment Outpaces Containment Capabilities

AI agent security machine identity enterprise AI deployment cybersecurity governance agentic AI

Enterprises are rapidly deploying AI agents, creating a new and complex cybersecurity frontier. Research from Dataiku and Harris Poll highlights that the UAE is a global leader in this deployment, with 62% of CIOs reporting more than 50 AI agents in production and 15% reporting over 500. However, a significant containment gap exists: only 5% of UAE respondents can reliably identify and contain a problematic agent within one to two hours.

While 80% of UAE CIOs have encountered agents that violate business intent, most incidents are detected before causing financial or operational impact. Despite this, the rise of autonomous agents is forcing a shift in security models. Experts argue that traditional Privileged Access Management (PAM) is insufficient for machine identities that operate at scale and speed. Companies like NVIDIA and CrowdStrike are collaborating on full-stack security platforms to provide independent controls outside the agent itself.

Security leaders face challenges in visibility and governance, with only 47% of CISOs expressing confidence in their ability to identify every AI agent in their environment. The industry is seeing increased consolidation and new specialized startups, such as Rig Security, aiming to map the relationships between agents and the resources they access to prevent unauthorized data exposure and identity-based attacks.

Entities

Amazon · BMW · Blueprint Alliance · CrowdStrike · Dataiku · Eric Kelleher · Nvidia · OKTA · Oracle Corp. · Palo Alto Networks · The Harris Poll · United Arab Emirates

Claims

What the coverage asserts, and how many sources carry each claim.

Sources

Time To Enforce Zero Trust On AI Agents [forbesliberia.com]
UAE Organisations Report Large-Scale AI Agent Deployment, but Containment Remains a Challenge [arabianreseller.com]
Reco raises $55M as AI agent security startups crowd the market [cybernoz.com]
Securing data at the source in the age of agentic AI [www.europesays.com]
Visa Strengthens AI Cybersecurity Defences as Risks from Autonomous Attacks Grow [influencermagazine.uk]
A Win for Defenders: CrowdStrike and NVIDIA Extend Security Across the AI Stack [massive.news]
Le RSSI d'Oracle place l'humain au centre de la gouvernance de la sécurité de l'IA [www.prodiris.fr]
AI agent chaos means ‘least privilege’ security isn’t enough anymore | The AI Journal [aijourn.com]
Giving AI Agents the Keys is Premature [www.b2bnn.com]
UAE Scales AI Agents but Faces Enterprise Containment Gap [enterpriseitworldmea.com]
Register Now: The Questions CISOs Should Be Asking About Agentic AI - A Live Panel [quisitive.com]
AI agent identity security demands layered defenses, Omdia says #AI - National Cyber Security Consulting [nationalcybersecurity.com]
Rig Security emerges from stealth with $12M to tackle AI identity risks [cryptobriefing.com]
Astra, la vista, baby. [thecyberwire.com]
UAE firms lead the world in deploying AI agents, but most can't contain a rogue one quickly, survey finds [gulfnews.com]
Agentic security strategy faces a shrinking breach window- SiliconANGLE [siliconangle.com]
SPONSORED: Close the observability gap with agentic observability [www.theregister.com]
Yes, You Can Change AI's Opinion. Here's How. [www.searchenginejournal.com]
Reco raises $55M as AI agent security startups crowd the market [techcrunch.com]