Thales Survey Finds AI Deployment Outpaces Consumer Trust
Thales's 2026 Digital Trust Index found that 93% of 200 surveyed IT decision-makers were already using, deploying or planning AI initiatives, while 23% of consumers trusted companies to use AI responsibly with their data. The research covered 14,300 consumers, 1,300 partner users and 200 IT decision-makers across 13 countries. The result makes transparent data handling and user control practical deployment requirements, not optional messaging.
Thales's 2026 Digital Trust Index found a sharp contrast between organizational AI adoption and consumer confidence. Among 200 surveyed IT decision-makers, 93% said they were already using, deploying or planning AI initiatives. Among the consumer respondents, only 23% said they trusted companies to use AI responsibly with their data.
The study was conducted by Vanson Bourne in January and February 2026. It surveyed 14,300 consumers, 1,300 business-partner users and 200 IT decision-makers across 13 countries. That methodology is important: the two headline percentages describe different respondent groups, so the 70-point difference is a contrast rather than a measure of individual attitudes changing over time.
Control and visibility shape trust
The official Thales release provides more detail than the July 24 commentary that surfaced the figures again. It says 77% of consumers were concerned about AI agents acting on their behalf online, while only 16% clearly understood how companies collect and use their personal data. Consumers also associated visible security controls with confidence: 69% said multifactor authentication increased trust and 68% said the same about passkeys.
Security Affairs connected the findings to the type and sensitivity of an AI task. Consumers were more receptive to AI used for security or low-stakes assistance than to autonomous systems making financial, travel or purchasing decisions. That does not establish a universal acceptance threshold, but it shows why a single organization-wide "AI enabled" label is too broad to predict user response.
What deployment teams should measure
For product, data and security teams, the practical question is not whether a model is present. It is whether users can tell what the system does, which data it receives, how long that data is retained, and when a person can intervene. Those controls should be testable in the product rather than left to a policy page.
A deployment review can translate the survey's trust concerns into concrete checks: disclose the AI-assisted step at the point of use, minimize and label data access, obtain permission before consequential actions, provide a human escalation path, and record whether users understand and use those controls. Teams should measure abandonment, opt-outs, overrides and complaints by use case instead of assuming that a high enterprise adoption rate implies consumer acceptance.
The index is vendor-sponsored survey research, not evidence that every market or product will reproduce the same percentages. Its value is the scale of the sample and the operational distinction it draws between adoption, transparency, security and autonomous action.
Key Points
- 1Thales found that 93% of 200 surveyed IT decision-makers were using, deploying or planning AI, while 23% of consumers trusted companies to use AI responsibly with their data.
- 2The January-February 2026 study surveyed 14,300 consumers, 1,300 partner users and 200 IT decision-makers across 13 countries.
- 3Deployment teams can address the reported trust concerns with use-case-specific disclosure, limited data access, permission gates, human escalation and measured user outcomes.
Scoring Rationale
The large cross-country survey gives product and security teams useful evidence about AI adoption, data trust and autonomous-action concerns, but it is vendor-sponsored perception research rather than a causal or product-performance study.
Sources
Primary source and supporting public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
