benchmarks

Research

New Benchmarks Expose Security Gaps in Enterprise AI Agent Access Controls
Byswgoettelman May 8, 2026

New research exposes a dangerous AI agent failure mode: when access controls silently filter restricted data, agents give confident but materially incomplete answers. A new 72-task benchmark measures the risk.

Read More New Benchmarks Expose Security Gaps in Enterprise AI Agent Access Controls
Research

Small AI Models Match GPT-5 on Routine Agent Tasks, Study Finds
Byswgoettelman May 5, 2026

New AgentFloor benchmark finds small open-weight AI models can match GPT-5 on routine agent tasks — suggesting enterprises could slash costs by routing most calls to smaller models.

Read More Small AI Models Match GPT-5 on Routine Agent Tasks, Study Finds