F-Secure’s new consumer survey highlights a disconnect between how people perceive the risks of AI and how they use the technology. While nearly 90% of AI users believe they are responsible for verifying AI-generated answers, 70% say they check those responses only sometimes, rarely, or never. The findings raise concerns as consumers increasingly rely on AI for personal decisions involving health, finances, relationships, and other important areas.
The risks aren’t theoretical; they’re already showing up in how people use tools like ChatGPT, Claude, and Gemini. Ironically, the people most exposed to AI’s mistakes are often the ones doing the least to catch them. F-Secure’s survey of 1,500 AI users in the US and UK digs into why (and what it’s costing them). Among the findings:
- Nearly half (47.3%) of AI users say they stop questioning an answer if it sounds right. That’s more than twice as common as any other reason given for not verifying a response.
- One in four (25.2%) have followed AI advice on a major real-life decision. Proof that users aren’t just casually using chatbots for casual questions or hypothetical scenarios.
- More than one-third (35.3%) have used AI to fact-check AI. Asking another AI chatbot to verify an answer can give people a false sense that they’ve independently checked it, when they’re really just asking AI to check its own work.
- Frequent AI use is associated with a lower bar for accepting answers. Among people who use AI for many different types of tasks, 59.7% accepted answers without further verification, compared with 28.6% among those who only use AI for low-stakes questions.
- Non-users want protection from AI-related risks before considering using them. Among people who don’t currently use AI chatbots, 46.1% said detecting AI-generated scams or manipulation would be the feature most likely to change their mind into using it, and 45% say their priority is AI preventing their data from being stored or shared.
The findings are clear: Users don’t really trust AI, but they still regularly use its answers without checking them first – despite saying they’re concerned about inaccuracy. And 83.9% of respondents have multiple concerns about AI, but still use it at least weekly.
“I don’t think the fix is asking people to be more careful. Users already believe checking the accuracy of the content is somewhat their responsibility,” said Amel Bourdocen, Senior User Researcher at F-Secure. “That responsibility sits with the people building these tools, not the people using them. A system that can sound confident regardless of whether it’s right should be the one required to show its uncertainty, not the user.”
To learn how F-Secure is helping close the AI trust gap, visit https://www.f-secure.com/en/partners/illuminate-research/consumer-ai-survey-april-2026
Related News:
Nile: The State of Networking, Security & AI in Financial Services
Darktrace Launches SECURE AI for Enterprise AI Security and Governance
