US firms report more AI-related hiring errors than UK

US companies are reporting significant AI-related errors at more than twice the rate of UK firms, according to a TestGorilla survey of senior hiring leaders in the two markets.

The research found that 33% of US organisations said an employee's over-reliance on AI had led to a significant error in the past six months, compared with 13% in the UK. Across both markets, 59% said they had made a bad AI hire in the past year, defined as someone who could discuss AI tools and workflows during recruitment but could not apply that knowledge effectively once hired.

The data suggests a difference in how employers assess AI fluency before making an offer. In the US, 45% of employers set the minimum bar at basic tool awareness, meaning candidates only need to know which tools exist and where they may be used. In the UK, that figure was 29%.

British employers were more likely to require candidates to show they could use AI independently and verify the results before being hired. Yet only 26% of organisations across both countries formally require candidates to complete that kind of test during the hiring process.

The findings suggest businesses are placing growing weight on AI familiarity even as many have yet to settle on a clear standard for what that means in practice. More than half of hiring managers, 53%, said they now prioritise AI fluency over domain expertise, making it the most common hiring priority across the two markets.

Hiring gap

The survey also highlighted that some employers still do not believe they need a formal definition of AI fluency. Among organisations that had not defined the term, 29% in the US said no definition was necessary, compared with 22% in the UK.

That matters because employers are trying to judge a skill that is becoming harder to spot through conventional recruitment methods. As applicants increasingly use AI tools to write CVs, refine applications and rehearse interview answers, recruiters may find it harder to distinguish between candidates who understand how to work with the technology and those who can simply talk about it confidently.

"Hiring leaders I speak with keep coming back to the same thing: the process was built for a different era, designed to find people who can describe their work well. For AI fluency, that is no longer enough," said Wouter Durville, Chief Executive Officer and Co-Founder of TestGorilla.

He said employers that are reducing the gap are changing the evidence they ask candidates to provide. "The organizations closing this gap are the ones willing to change what they ask candidates to show them, not just tell them," Durville said.

Changing signals

The survey covered 1,928 senior hiring leaders in the US and UK across 29 industries. It examined how organisations define and measure AI fluency and included employers hiring between one and more than 250 roles a year.

The figures add to a broader recruitment debate over whether traditional screening methods are becoming less reliable when candidates can use generative AI to improve written communication and interview preparation. For employers, the challenge is not access to AI tools but whether internal hiring processes test practical judgement, verification and independent use.

Hung Lee, a 25-year recruiting industry veteran and curator of the talent industry newsletter Recruiting Brainfood, said the spread of AI is making candidate assessments look increasingly similar. "The 59% bad hire rate does not surprise me," he said.

"AI is homogenizing how candidates present themselves. The CVs, the applications, the interview answers are starting to look the same. Your ability to identify who is genuinely AI-fluent is actually decreasing as more AI enters the process," Lee added.

Another recruitment executive said the problem starts with unclear expectations at the hiring stage. "The question most hiring managers never ask is: what does AI fluency actually look like in this role, and how do I know it when I see it?" said Lou Adler, Chief Executive Officer of Performance-Based Hiring.

"Until managers can answer that, they are not screening for AI fluency. They are screening for people who can talk about it convincingly," Adler added.