AI agent security risk is rising fast, but are controls keeping up?

TL;DR: A survey of 300 enterprise leaders found 97% expect a material AI-agent-driven security or fraud incident within 12 months, with nearly half expecting one within six months, while only 6% of security budgets are allocated to the risk, according to Arkose Labs. The gap is now governance, visibility, and attribution, because autonomous access can move faster than review cycles can respond.

NHIMG editorial — based on content published by Arkose Labs: AI 97% of Enterprises Expect a Major AI Agent Security Incident Within the Year

By the numbers:

• 97% of respondents expect a material AI-agent-driven security or fraud incident within the next 12 months.
• Only 6% of security budgets are currently allocated to this risk.
• The report is based on a global survey of 300 enterprise leaders across security, fraud, identity and AI functions.

Questions worth separating out

Q: How should security teams govern AI agents that operate with legitimate enterprise credentials?

A: Security teams should govern AI agents as non-human identities with explicit ownership, traceable scope, and monitored action paths.

Q: Why do AI agents create more attribution risk than conventional automation?

A: AI agents create more attribution risk because they can choose actions at runtime and move across multiple services using valid credentials.

Q: What do organisations get wrong about AI agent governance?

A: Organisations often treat AI agent governance as a policy exercise after deployment instead of an identity problem before deployment.

Practitioner guidance

• Inventory every AI agent identity and its sponsor Create a register of service accounts, API tokens, and application identities used by AI agents, then assign a human owner and a business purpose to each one.
• Correlate agent actions across systems and APIs Instrument logs so that one agent session can be traced across credential use, data access, and downstream actions in connected services.
• Bound automated decision chains before production Require explicit scope boundaries for what an agent may retrieve, trigger, or disclose, then review those boundaries before release into live environments.

What's in the full report

Arkose Labs' full report covers the operational detail this post intentionally leaves for the source:

• Regional breakdowns across North America, Europe, and Asia-Pacific that help you compare risk by operating model.
• Function-level survey results across security, fraud, identity, and AI teams for internal benchmarking.
• The report's full action guide for building visibility, attribution, and classification capabilities around agentic AI.
• The statistical methodology, including the 95% confidence level and ±5.6% margin of error.

👉 Read Arkose Labs' 2026 report on AI agent security and enterprise readiness →

AI agent security risk is rising fast, but are controls keeping up?

Explore further

View Full Forum →  |  NHI Foundation Course →