AI and Labour Relations: Employers Expect Organising, Few Feel Ready

By Esther Smith

On AI and labour relations, fewer than one in ten employers feel very prepared. Nearly nine in ten expect unions to use worker fears about AI to drive organising over the next 12 months. Employers can see the problem coming, and most have yet to plan for it.

Both figures come from Littler’s 2026 Labor Survey Report, published on 29 September 2026. Littler is an employment law firm that advises employers. It surveyed 665 in-house lawyers, HR professionals and business executives, mostly in the US, at unionised and non-unionised organisations. The survey records what employers expect. It says nothing about what unions will do.

AI and labour relations planning trails the expectation of organising

Planning for AI and labour relations shows the shortfall. Only 29% of employers have updated their labour relations strategy since the start of 2025. Around four in five include employee engagement and legal compliance in that strategy. Fewer than one in three include an AI strategy.

Engagement is where the effort has gone. Some 96% of employers say they worked on employee engagement over the past year, yet only 17% communicate proactively and consistently about what AI means for jobs. Feedback forums and culture programmes collect how people feel, which is a different job from telling people what is happening to their own role. Pulse surveys can also miss the behavioural effect of AI, as EMex’s reporting on AI attribution and discretionary effort showed.

Unionised employers show the same pattern. Most have a collective bargaining strategy, yet fewer than four in ten have a contingency plan for a strike or work stoppage.

Littler also found that 30% of employers have seen staff turn more vocal or disruptive on social media since early 2025, and 24% in public forums.

The shortfall shows up in change planning generally, too, with global research finding organisations underprepared for change pointing the same way.

What turns worry about AI into something organisable

The headline figures stop at the 87%. The useful question sits underneath it: what gives a vague worry a shape that organisers can use?

Usually the order of events does. A tool gets approved by technology and finance. Output targets rise a quarter later. An entry-level role changes shape. Nobody in the affected team was consulted, the explanation arrives after the rumour, and the timeline comes last or never. Staff then compare notes across teams, and the comparison supplies the grievance.

Two thirds expect AI to reshape job responsibilities, and three in ten expect fewer entry-level roles. Staff will feel those changes first.

Jonathan Levine, co-chair of Littler’s Labor Management Relations Practice Group, says some employers treat AI as “an IT project when it’s really a workplace change project”. That’s a lawyer’s framing, and a fair one. An IT project has a go-live date. A workplace change has a consultation period.

Littler’s own view is that employers who answer the “how will this affect my job” question early can reduce labour relations risk. The logic is still easy to test inside your own organisation by checking how early the affected staff heard.

The case for reading the 87% with caution

There’s a reasonable argument that the figure overstates the risk. Expecting an organising attempt is a weaker claim than forecasting a result. A survey of lawyers, HR leaders and executives will also over-represent people whose job is to spot risk. The 9% counts only the top answer, so many employers who say they’re somewhat prepared may be placed well enough.

Even on the cautious reading, the 87% records what employers themselves think carries the most pull with staff: job insecurity. They control more of what feeds it than any union does, which makes AI and labour relations a planning question long before it becomes a bargaining one.

The union figure is specific to the US. UK and European employers work under different rules on collective representation and consultation, so the organising mechanics won’t transfer directly. The pattern underneath does, and it applies to employers with no union at all. Staff who aren’t told what AI means for their role will form their own view.

Employers also don’t need every answer to start explaining. EMex has covered why employees don’t need leaders with all the answers, and a plain account of what’s known, what’s being tested and when more detail will follow keeps credibility intact.

What sits within your control

Worth a look next time an AI tool reaches sign-off: who in the affected team was consulted before the decision, what has been explained in plain terms about tasks, targets and reporting lines, and when staff will hear more. On AI and labour relations, those three answers sit with HR and people leaders rather than IT, and they’re settled long before a union asks anything.

An employee correction window after rollout gives the third answer a date. A consultation step gives the first one a name, and an explanation written for the people affected gives the second one an audience.