The verification gap is the distance between what a company says its AI agents do
and what it can show. It is the first question a customer, an auditor or a regulator will
ask. Close to half the companies mentioning agents have no public answer.
~12,000
company websites analyzed
771
publicly mention AI agents
43%
show no public verification language
shows verification language (439)
silent on verification (332)
The gap, made visible.
Each dot represents one company that publicly mentions AI agents. Green dots represent companies whose public website language discusses monitoring, evaluation, governance or human oversight.
What the research found
01
AI signals are widespread. AI-native depth is rare.
A third of organizations show AI signals on their websites. Only 0.7% appear AI-native.
02
The agent wave is already visible.
771 companies publicly mention AI agents or agentic workflows.
03
43% are silent on verification.
332 of 771 companies with agent signals show no public language about monitoring, evaluation, governance or human oversight.
04
Governance language is ahead of evaluation.
32% mention governance and 32% mention human oversight, while only 25% mention evaluation.
05
Accountable sectors show less public verification language.
90 regulated-sector companies mention agents with no public verification language.
See the full picture
The Verification Gap 2026 breaks the findings down by AI maturity, industry and geography, with a closer look at how companies talk about autonomy, evaluation, governance and human oversight.
See what close to 12,000 company websites reveal about AI agent readiness and where the gap between agent claims and public evidence is widest.
