Americans are using AI, but they still don’t trust it
Americans are using AI, but they still don’t trust it
Published on
30 December 2025
Chun Hui Lee image
Chun Hui Lee
Insights Writer

Adoption of AI is rising, but the fear factor is not budging. Latest results from Conjointly's AI Adoption and Opinion Tracker reveal Americans remain wary of the AI tools they use every day.


Conjointly launched the AI Adoption and Opinion Tracker in June 2023 to monitor how US consumers use and view AI technology. Between June 2023 and December 2025, Conjointly conducted 14 waves of research, surveyed approximately 500 Americans per wave for a total sample of 7,011. Each sample was weighted to match US age and gender demographics.

Fieldwork dates and sample sizes for each wave are as follows:

WaveFieldwork datesSample size
Wave 126 to 28 June 2023498
Wave 226 to 28 August 2023488
Wave 313 to 15 December 2023490
Wave 414 to 19 February 2024501
Wave 519 to 22 April 2024494
Wave 628 June to 30 June 2024494
Wave 727 to 29 August 2024495
Wave 89 to 11 November 2024486
Wave 99 to 15 January 2025540
Wave 1011 to 13 March 2025520
Wave 1111 to 13 June 2025497
Wave 1212 to 21 August 2025500
Wave 1314 to 20 October 2025509
Wave 1413 to 19 Devember 2025499

Rising consumer adoption usually suggests growing comfort with new technology. However, the tracking data tells a different story. Over 30 months, a paradox of necessity emerged where Americans increasingly integrate AI into their lives, yet their desire to halt its development shows no signs of weakening.

This article highlights findings across all 14 waves and identifies how both AI adoption patterns and attitudes toward the technology have evolved over this period.

ChatGPT adoption doubled over 30 months

Measured purely by usage metrics, AI has achieved remarkable penetration. Frequent ChatGPT usage more than doubled across the tracking period, climbing from 11% in Wave 1 (June 2023) to 23% in Wave 14 (December 2025). Trial users, i.e. those who have tried it once or a few times also grew from 18% to 31%.

Are you familiar with ChatGPT?
I use it frequently
Used once/few times
Heard/Not used
Never heard

The proportion who had never heard of ChatGPT dropped from 26% to 10%. This pattern held across all age groups, with each demographic showing a clear downward trend despite wave-to-wave variability. Those aged 65 and over saw the sharpest decline, dropping from 31% to 5%. The 30-44 age group followed, falling from 25% to 10%.

Are you familiar with ChatGPT? Percentage responding "Never heard of ChatGPT" by age group
18-29 years old
30-44 years old
45-64 years old
65 years old above

However, this surge in adoption and awareness failed to dampen public demand for a safety pause.

The “Safety Pause” sentiment remains stubborn

Agreement with the statement that “governments must implement a pause on advancing AI capabilities” tracked consistently above 55% throughout the tracking period. Support started at 64% in June 2023, fluctuated within a narrow band across 14 research waves, and stood at 59% by December 2025.

On a scale of 1 to 5, how much do you disagree or agree that governments must implement a pause on advancing artificial intelligence (AI) capabilities, at least until there are solutions to make AI development safe?
Agree
Strongly agree

ChatGPT usage makes little difference to pause sentiment over time. Across 14 waves, non-users averaged 61% agreement while users averaged 59%, both groups maintained support for a development pause throughout the tracking period, reflecting deeper concerns across multiple dimensions.

On a scale of 1 to 5, how much do you disagree or agree that governments must implement a pause on advancing artificial intelligence (AI) capabilities, at least until there are solutions to make AI development safe?
Those who never used or heard of ChatGPT
Those who frequently use or have trialled ChatGPT

Trust in AI software showed no upward trend

Agreement that software that uses AI is trustworthy ranged between 26% and 35% across all waves, showing no upward trend since the end of 2023.

On a scale of 1 to 5, how much do you disagree or agree that software that uses artificial intelligence (AI) is trustworthy?
Agree
Strongly agree

A persistent trust gap between ChatGPT users and non-users has tracked throughout the study period. ChatGPT users consistently showed higher trust levels, averaging 44% compared to 20% among non-users across the 14 waves.

Notably, the gap narrowed in the October 2025 and December 2025 waves, as user trust declined compared to previous waves while non-user trust remained at similar levels.

On a scale of 1 to 5, how much do you disagree or agree that software that uses artificial intelligence (AI) is trustworthy?
Those who never used or heard of ChatGPT
Those who frequently use or have trialled ChatGPT

Government regulation consistently viewed as inadequate

The persistent demand for a pause becomes clearer when examining how Americans view government oversight.

Throughout the tracking period, only one in four Americans believed the US government adequately regulates the AI industry. This perception of regulatory insufficiency remained relatively stable across all 14 waves, ranging between 19% and 28%, with most measurements clustering near 25%.

On a scale of 1 to 5, how much do you disagree or agree that the US government is doing enough to regulate the artificial intelligence (AI) industry?
Agree
Strongly agree

The 2024 election appears to have temporarily shifted perceptions of AI regulation. Republican confidence in government AI regulation climbed through the first half of 2025, rising from 25% in November 2024 to peak at 37% in June. From there, a reversal took hold. Republican agreement dropped to 33% in August, then fell sharply to 22% in October before recovering to 28% by December.

Democrats showed a different pattern. Their agreement fell from 29% in November 2024 to 23% in January 2025, then recovered gradually through mid-2025 before declining again in the final months.

By December 2025, Republican agreement had increased slightly to 28% compared to 25% a year earlier, while Democratic agreement dropped to 22% from 29%.

On a scale of 1 to 5, how much do you disagree or agree that the US government is doing enough to regulate the artificial intelligence (AI) industry?
Republican
Democrats

Fears about losing control stayed widespread

Nearly half of Americans agree that advanced AI systems cannot be switched off if they don’t want to be disabled. Agreement ranged between 44% and 56% across all 14 waves, showing no signs of diminishing despite growing familiarity with AI technology.

On a scale of 1 to 5, how much do you disagree or agree that you cannot simply switch off a sufficiently advanced artificial intelligence (AI) system if it does not want to be switched off?
Agree
Strongly agree

Age shapes AI autonomy fears dramatically. Young adults aged 18-29 were more than twice as likely as those 65 and over to believe advanced AI systems could resist shutdown, averaging 53% versus 21% across all waves. This generational gap persisted consistently throughout the 30-month tracking period.

On a scale of 1 to 5, how much do you disagree or agree that you cannot simply switch off a sufficiently advanced artificial intelligence (AI) system if it does not want to be switched off?
18-29 years old
30-44 years old
45-64 years old
65 years old above

Majority support for creator compensation

Since June 2023, at least six in ten Americans supported compensating artists and content creators when their work trains AI models.

While there has been a modest softening from the peak of 73% in late 2023 to 63% in both October and December 2025, support has remained above 60% in every wave measured over the 30-month period.

This sustained level of agreement suggests that while Americans grapple with questions of AI safety and governance, they remain consistent in their view that creators deserve payment when their work contributes to AI development.

On a scale of 1 to 5, how much do you disagree or agree that artists and content creators should be paid if their work is used in training artificial intelligence (AI) models?
Agree
Strongly agree

How to keep tabs on consumer adoption and attitudes

The persistence of AI concerns despite doubling usage rates, demonstrates that American attitudes toward AI are not evolving in predictable patterns. Trust is not building with adoption, while fear is not diminishing with familiarity.

These insights emerge only through continuous measurement over time. Tracker studies make this possible by surveying the same questions at chosen intervals, revealing how consumer preference and attitudes are shifting or stabilising.

For organisations navigating rapidly evolving markets, understanding these patterns makes the difference between reacting to temporary fluctuations and responding to genuine shifts in consumer sentiment. To explore how tracking studies can illuminate the trends shaping your product category, schedule a consultation with Conjointly’s research team.


Read these articles next:

Understanding Brand Preference through Market Research

Understanding Brand Preference through Market Research

This article examines the importance of establishing brand preference and brand equity and explores how market research helps businesses shape their brand to fit their target audiences.

View article
Stravito and Conjointly Work Together to Maximise Insight Value

Stravito and Conjointly Work Together to Maximise Insight Value

Conjointly is collaborating with Stravito, a flexible and user-friendly insights platform, to provide organisations with powerful tools to conduct research and share insights with teams worldwide.

View article
Conjointly achieves ISO 27701:2019 and ISO 27001:2022 certification

Conjointly achieves ISO 27701:2019 and ISO 27001:2022 certification

Conjointly, an all-in-one survey research platform, today announced that it has achieved ISO 27701:2019 certification for Privacy Information Management and ISO 27001:2022 certification for Information Security Management.

View article