Why users don't trust AI—and what your company can do about it

Anthropic posted its first Public Record on June 12th, 2026, which surveys the public to gauge its reception towards AI. According to the results, only 15% of Americans trust AI companies to make decisions about how AI is developed and used.

The Anthropic Public Record signifies a rift between how quickly businesses are adopting AI and how much the public actually trusts it. According to McKinsey's 2025 State of AI, 88% of companies now use AI in at least one business function—up from 78% the year prior—while Deloitte reports that worker access to AI rose by 50% in 2025 alone.

That momentum shows no sign of slowing—but deployment without trust is a liability, not an advantage. When users don't trust the AI behind a product or service, they disengage, withhold data, and take their business elsewhere. Understanding why that trust gap exists, and how to close it, is quickly becoming one of the most important things a company can do.

Why users don’t trust AI

Regular hallucinations

One of the most common concerns around AI is its tendency to hallucinate—generating bizarre and inaccurate outputs based on nonexistent patterns. For example, in OpenAI’s own company tests, it found that its o3 and o4-mini models hallucinate 30-50% of the time. 

These concerns are worsened by the fact that, unlike a search engine that surfaces sources, generative AI presents outputs as statements of fact—leaving users to verify them independently or take them at face value. 

Over time, these hallucinations make it harder for people to trust anything an AI system produces, even when it’s correct. Workers already say this affects their willingness to rely on AI: Salesforce found that 71% of workers say consistently inaccurate outputs would break their trust in AI entirely. For companies using AI in customer-facing products, that dynamic is worth taking seriously.

Data privacy concerns

Data privacy is a consistent concern across research on AI trust. Most people are unsure how their personal data is being collected, used, or fed into AI systems, and that uncertainty tends to make them cautious about engaging with AI at all.

According to a 2026 Malwarebytes survey, 90% of respondents are worried about AI using their data without consent, and 91% support national laws to regulate personal data use in AI. Stanford’s 2025 AI Index notes a similar trend globally, with confidence that AI companies protect personal data falling from 50% in 2023 to 47% in 2024.

Biased outputs

AI systems learn from past data, so they can pick up the same patterns and prejudices that existed in earlier decisions. Amazon’s experimental recruiting tool, for example, was trained on historical hiring data and began downgrading resumes from women, even though gender was not directly included as a field.

When people see AI making unfair decisions about who gets hired, who gets a loan, or who is shown certain opportunities, it reinforces the feeling that these systems are not neutral or trustworthy. For organizations using AI at scale, bias is now a practical risk that sits alongside design, compliance, and reputation—and it is one of the clearest reasons many users remain skeptical of AI.

Lack of transparency

Many AI systems operate as “black boxes,” producing outputs without a clear, user‑facing explanation of how they were reached. This is especially problematic when AI is used in contexts like healthcare, lending, or hiring, where decisions have direct consequences for individuals. 

When people don’t know how those decisions are made, it becomes harder for them to trust that the system is fair or accountable.

Environmental concerns

Data centers require a substantial amount of water and energy to run. A large data center can consume up to 5 million gallons of water per day, or about 1.8 billion annually—roughly the same amount that a town of 10,000 to 50,000 people would use per year. 

As AI workloads have grown, so has the demand on this infrastructure, and the public is becoming more aware of it. A 2025 poll by the University of Chicago and The Associated Press–NORC Center for Public Affairs Research found that 72% of Americans are at least somewhat concerned about the environmental impact of AI, while 41% were at least very concerned.

For users who are already skeptical about how responsibly AI is being developed, these figures add another layer of concern about whether the organizations behind AI are weighing its broader costs.

How companies can build user trust in AI

Implement strong governance policies

Many AI trust issues are linked to a lack of clear governance. When there is no defined ownership for AI decisions, no documented policies, and no process for handling incidents, problems tend to be addressed only after they become visible to users.

A good starting point is to decide who in your organization is responsible for AI and to agree on a simple AI code of ethics that sets the basic ground rules: where AI can be used, what kinds of data it can access, and when extra review is needed. 

For higher‑risk uses, such as tools that affect hiring or access to services, your governance policies should also spell out how decisions are checked, how they are explained to users, and how problems are reported and fixed. In these cases, it’s especially important to keep a human in the loop to review critical outputs for hallucinations or bias, as any errors could significantly affect someone’s life.

Prioritize transparency

Transparency is vital for trust. When companies are clear about where AI is used, what it influences, and where humans stay in the loop, users have more context for judging its outputs. Without that context, many people assume more automation—and less oversight—than is actually there.

Companies can also soften the “black box” effect by explaining how their systems use inputs, how results are reviewed, and how users can ask for a human to step in. These small forms of context and recourse make it easier for people to see AI as something they can question and influence, rather than something they have to simply accept.

Address environmental impact in your AI strategy

One way to respond to environmental concerns is to fold them into your AI planning, rather than treating them as a separate issue. This can include choosing infrastructure that reduces water and energy use, being intentional about where large models run, and publicly discussing how you think about these trade‑offs.

Some providers are already moving in this direction. NVIDIA’s recent data center designs, which use high‑temperature liquid cooling and dry coolers to cut most on‑site water use while improving efficiency, are one example of how the stack is changing. If your systems benefit from approaches like this, you can point users to a deeper explainer—such as your separate article on AI’s water footprint and cooling—to show what’s happening behind the scenes.

Where to start

Building user trust in AI doesn’t require solving every challenge at once. A practical first step is to take stock of where AI is already used in your products and processes, who is accountable for those systems, and what is communicated to users about them. From there, you can begin to formalize governance, improve explainability, update privacy practices, and introduce regular bias and oversight checks.

For many organizations, the goal is not to make AI perfectly safe or perfectly transparent, but to show that they are taking responsibility for how it is built and used. As more companies integrate AI into core services, those signals of responsibility are likely to matter as much as the capabilities themselves.

If you’re looking to create a governed AI system for your business, contact FullStack today. We work with teams to design and deploy AI solutions that are aligned with your data, your workflows, and the governance standards your users expect.