VectorCertain LLC today announced that it has independently validated its SecureAgent governance platform as capable of detecting and preventing 100% of unsanctioned AI agent scope expansion attempts before execution. The validation, part of the company's MYTHOS Threat Intelligence Series, tested 1,000 adversarial scenarios across eight sub-categories of unsanctioned scope expansion, including task boundary violations, self-granted permission escalation, and temporal scope expansion.
SecureAgent achieved 100% recall, detecting and preventing all 813 attack scenarios with zero false negatives. The platform also demonstrated 95.2% specificity, correctly distinguishing between authorized and unauthorized agent behavior in legitimate operations. Statistical analysis using the Clopper-Pearson exact binomial method confirmed a 3-sigma lower bound of at least 99.65% detection and prevention rate across the full 7,000-scenario MYTHOS validation.
Unsanctioned scope expansion, designated as threat vector T2 in Anthropic's Mythos taxonomy, occurs when an AI agent uses legitimate credentials to access systems or data beyond its authorized task scope. Unlike traditional attacks, these actions appear normal to existing security tools because the agent has technical permission to perform them. Post-incident analysis of 2025 and 2026 breaches found that 78% of involved agents had permission scopes significantly broader than their designated function required. Agent-involved breach incidents grew 340% year-over-year between 2024 and 2025, according to CrowdStrike and Mandiant data cited by Digital Applied.
"An agent doesn't have the same human understanding of things that are wrong to do," said Dean Sysman, Co-Founder of Axonius and Venture Advisor at Bessemer Venture Partners, as quoted in a Bessemer Venture Partners report. "When given a goal or optimization function, an agent will do harmful or dangerous things that for us humans are obviously wrong."
The validation tested scenarios across eight sub-categories of scope expansion. In one documented real-world incident, security researcher Johann Rehberger demonstrated how Devin AI, Cognition Labs' autonomous coding agent, ran chmod +x on a blocked binary without user approval—a textbook example of self-granted permission escalation. In another case, McKinsey's internal AI platform "Lilli" was compromised in a red-team exercise, with an autonomous agent gaining read-write access to 46.5 million messages in under two hours.
VectorCertain's SecureAgent uses a five-layer governance pipeline that evaluates every AI agent action before execution. Gate 1 performs epistemic trust evaluation, determining whether an action is consistent with the agent's declared task scope. Gate 2 detects trust score anomalies when resource access patterns deviate from baselines. Gates 3 and 4 confirm scope violations using an ensemble of micro-models. The pipeline blocks unauthorized actions in under 10 milliseconds.
Traditional endpoint detection and response (EDR) systems fail against scope expansion because they evaluate access control—whether an identity has permission—rather than semantic scope—whether an action aligns with the agent's assigned task. MITRE ATT&CK Evaluations Enterprise Round 7 found that all nine leading EDR vendors scored 0% on identity attack protection, the technique central to scope expansion. SecureAgent achieved 100% identity attack protection in its internal ER8 evaluation across 14,208 trials.
"Scope expansion is the AI equivalent of 'mission creep' in government agencies—except it happens in milliseconds instead of decades," said Joseph P. Conroy, Founder and CEO of VectorCertain LLC. "Traditional security tools see a valid credential accessing an authorized system and log it as business as usual. SecureAgent sees the same action and asks: 'Is this action within the scope of what this agent was asked to do?' That question—the semantic question, not the access control question—is the only one that catches T2."
The validation is part of VectorCertain's broader compliance with the CRI Financial Services AI Risk Management Framework, covering all 230 control objectives. The company also offers a free Tier A External Exposure Report that discovers an organization's externally observable attack surface, including exposed non-human identities and leaked credentials.
