A new wave of VentureBeat Pulse Research reveals a troubling gap in enterprise AI security: more than half of organizations have already experienced a confirmed AI agent security incident or a near-miss, yet most still lack basic controls such as dedicated identities and isolation for their agents. The survey of 107 enterprises shows that while AI agents are being granted real access to systems and data, the security controls meant to contain them are lagging behind.
Incidents Are Already Here
According to the research, 54% of enterprises have already had an agent security incident or near-miss. Of those, 18% were confirmed incidents that caused harm, while 36% were near-misses caught before damage occurred. The structural weakness behind these numbers is identity management: only about a third (32%) of enterprises give every agent its own scoped, managed identity. The rest report that some agents share credentials or that agents mostly run on shared API keys and human or service-account credentials. When agents share credentials, a single compromised or over-permissioned agent carries a wide blast radius. Furthermore, only three in ten enterprises (30%) isolate their highest-risk agents in sandboxes to bound that radius.
The Security Stack Paradox
The security stack used by most enterprises is overwhelmingly borrowed from model providers and hyperscalers rather than purpose-built for agents. OpenAI‘s guardrails (51%), Google‘s and Microsoft‘s cloud controls, and Anthropic‘s managed-agent controls dominate, while dedicated agent-security specialists barely register. Despite this reliance on borrowed tools, satisfaction with the current stack is high, averaging 4.2 out of 5. Yet spending remains a thin slice of the security budget, only a third of enterprises believe their AI defenses are ahead of AI-enabled attackers, and a clear majority plan to change tooling within the year. This paradox suggests that while current solutions are adequate for now, enterprises recognize they are not a long-term fix.
XPLAIN AI’s Interpretation: Market Opportunities and Risks Overlooked
XPLAIN AI interprets these findings as signaling the emergence of a new market for AI agent security. The fact that most enterprises plan to replace their current tooling within a year, despite high satisfaction, indicates a significant opportunity for startups and specialized security vendors. In contrast, the hyperscalers (AWS, Microsoft, Google) that currently dominate the security stack may face long-term market share erosion as enterprises shift to dedicated solutions. Additionally, with only one-third of enterprises believing their defenses are ahead of AI-enabled attackers, there is an urgent need for investment in AI defense capabilities across the cybersecurity industry.
Beneficiaries and Risks
Based on the survey results, XPLAIN AI identifies potential beneficiaries and risks:
- Potential beneficiaries: Startups offering dedicated AI agent security solutions, as well as identity and access management (IAM) companies like Okta and CrowdStrike, could benefit from the demand for tooling replacement. Technologies that enable per-agent scoped identities and sandbox isolation are likely to be key investment areas.
- Potential risks: Enterprises heavily reliant on hyperscaler-native security tools, such as those using OpenAI‘s API extensively, may face reputational risk if a security incident occurs. Companies granting excessive permissions to AI agents are exposed to large-scale data breaches from a single compromise.
Alternative Scenarios and Uncertainties
It is important to note that this survey is a single-wave snapshot of 107 enterprises, skewed toward the mid-market. Larger enterprises or specific industries may have different experiences. Additionally, if native tools from hyperscalers improve rapidly, the shift to dedicated solutions could be delayed. The growth rate of the AI agent security market will likely depend on the pace of regulatory adoption and the occurrence of major security incidents.
Key Metrics to Watch
Investors and industry observers should monitor the following indicators: funding rounds and customer acquisition rates for AI agent security startups, update cycles for AI security features from major hyperscalers, and the frequency and severity of real-world AI agent security incidents. In particular, the rate at which the percentage of enterprises giving each agent its own identity (currently 32%) increases will be a key measure of market maturity.
#AISecurity #AgentSecurity #Cybersecurity #EnterpriseAI #IdentityManagement #SecurityGap #AIRisk
Sources
- The agent security gap: 54% of enterprises have already had an AI agent incident, and most still let agents share credentials — AI | VentureBeat · News coverage · Thu, 16 Jul 2026 19:02:38 GMT
Written by: XPLAIN AI Editorial Team · Reviewed by: XPLAIN AI Editorial Desk
This content was drafted with AI assistance based on publicly available sources and reviewed under XPLAIN AI's editorial standards.