Persistent Distrust: 9 Out of 10 Internet Users Still Doubt AI

Concerns About Personal Data

A survey conducted by Malwarebytes between late January and early February 2026 among 1,235 subscribers to its newsletter reveals a persistent distrust towards artificial intelligence (AI). Nine out of ten respondents express concern about the use of their data by AI. Although AI assistants are increasingly integrated into search engines, corporate messaging, and everyday applications, this omnipresence does not seem to bolster user trust.

The Ongoing Ambiguity of Privacy Policies

On social media, users generally understand the mechanisms of data collection through clicks and "likes." However, with AI, the uncertainty is greater. Interactions with AIs are often conversational and intimate, involving confidential meeting notes, medical questions, or personal dilemmas. Yet, it is unclear what happens to these exchanges. Users wonder if these conversations are stored, if they are used to train models, or if employees can access them.

Privacy policies exist, but few people read them before using a tool. However, 48% of respondents now say they consult them, an increase from 43% the previous year. This change coincides with the CNIL and ANSSI opening tests for their PANAME project, an open-source tool designed to verify whether an AI model has retained personal data from its training.

Distrust Translated into Concrete Actions

In the face of this opacity, internet users are not remaining passive. 88% of those surveyed avoid sharing personal information with tools like ChatGPT or Gemini. 84% refuse to provide health data to an AI. This distrust translates into concrete actions: 43% of users have stopped using ChatGPT and 42% have left Gemini. Distrust towards AI reflects a heightened skepticism towards tech companies, with 92% of respondents worried about the misuse of their data by private companies, up from 89% in 2025.

Increasingly Proactive Users

A year ago, 74% of respondents felt powerless regarding the dispersion of their data online. This figure has dropped to 63% in 2026, indicating a growing awareness. The use of VPNs has increased from 42% to 46%, and two-factor authentication is now adopted by 76% of users, compared to 69% the previous year. Additionally, 82% of respondents actively refuse data collection when possible, up from 75% in 2025. Finally, 25% of those surveyed use personal data deletion services, although these measures do not erase past traces, they limit new exposures.

These developments demonstrate an increased vigilance among internet users regarding the protection of their personal data, a crucial issue in a context where AI is becoming more integrated into our daily lives.