AI Agent Security Risks: Why 65% of Enterprises Experience Excessive Permission Failures
A new survey by Enterprise Management Associates (EMA) reveals significant gaps in enterprise control over AI agent permissions, with 65% of surveyed companies reporting instances where AI agents performed operations beyond their intended scope.
AsiaAI Publisher
·
September 5, 2026 ·
2 min read · Source: iThome · Issue #87
AI & Machine Learning
East Asian Technology Intelligence
Japan & China tech news — translated, contextualized, and delivered for Western readers.
Free. Unsubscribe anytime.
This story ran in Issue #87, alongside three other stories.
📊 Featured Chart
Source: EMA Survey of 202 Large Enterprise Tech/Security Execs
A new survey by Enterprise Management Associates (EMA) reveals significant gaps in enterprise control over AI agent permissions, with 65% of surveyed companies reporting instances where AI agents performed operations beyond their intended scope. While 94% of respondents were confident their AI agents did not exceed necessary permissions, only 32.7% actually implemented the principle of least privilege.
This report highlights a critical and growing concern for Western businesses adopting AI, particularly as AI agents move from testing environments into formal operations. The OWASP Top 10 for LLM security risks already lists ‘Excessive Agency’ as the third-highest threat, underscoring the universal nature of this challenge beyond East Asian markets.
