C
    collaboration.tech
    All insights
    Sep 23, 20269 min readDarek Ambroziak

    Three in Four Companies Haven't Told Employees Why They're Implementing AI. What Fills That Silence?

    Organizational AI integration surged from 38% to 47% in a year, while the share of employees who heard a clear AI plan stayed flat at 25% (Gallup, 2026). What grows in that silence — and the three elements a plan must contain to actually land.

    Illustration of a megaphone emitting empty dotted lines toward employees whispering among themselves — the silence left when no AI plan is communicated.

    In a single quarter, the share of U.S. employees whose organizations have integrated AI jumped six points to 47% — the sharpest move Gallup has recorded. The share who say their organization has communicated a clear AI plan hasn't moved in a year: it sits at 25% (Gallup, 2026). Implementation and message are measurably diverging, and the gap is where both shadow AI and resistance grow.

    TL;DR

    • Organizational AI integration, per U.S. employees: 38% → 38% → 41% → 47% across Q3 2025–Q2 2026 (Gallup, 2026).
    • Clear-plan communication over the same era: 15% (May 2024) → 22% → 24% → 26% → 25% → 25% (May 2026). Flat for a year while integration surged (Gallup, 2026).
    • 20% of employees don't know whether their organization has adopted AI at all — unchanged quarter after quarter (Gallup, 2026).
    • Using AI does not dissolve the fear; how people are treated does. Frequent users are more than twice as likely to fear job elimination, and feeling cared for and respected weakens that link within the same worker over time (Makridis, Gallup, 2026).
    • Silence is not neutral. In the absence of an answer to "why are we doing this," employees supply their own — and typically the worst available one.

    What do the two Gallup trend lines show when you put them side by side?

    The integration line moves like a technology curve: 38% of U.S. employees said their organization had integrated AI in Q3 2025, 38% in Q4, 41% in Q1 2026, and 47% in Q2 — with the last jump the sharpest in Gallup's measurement (Gallup, 2026). The communication line moves like a flat tax: asked whether their organization has communicated a clear plan or strategy for integrating AI, 15% said yes in May 2024, 22% in May 2025, then 24%, 26%, 25%, and 25% through May 2026 (Gallup, 2026).

    Put differently: for every hundred employees, integration gained nine points in a year while communication gained three, then stalled. Three in four employees now work for organizations that — by the employees' own account — have not explained what the AI push is for. Deployment has decoupled from meaning.

    Why is "we'll communicate once the pilot delivers results" an expensive mistake?

    Because organizational silence is not an empty channel — it is a channel that employees fill themselves. A definition worth writing down: in the absence of an official answer to "why is the company doing this," every employee constructs a private answer, and under job-security uncertainty the privately constructed answer skews pessimistic. Gallup's data gives that pessimism a number: 23% of employees in AI-adopting organizations consider it likely their job will be eliminated within five years due to AI or automation, versus 18% of all employees (Gallup, Q1 2026).

    The pilot-first logic assumes communication is a report on results. For employees it is something else: a signal about intent. Waiting six months for pilot metrics means six months in which the workforce's working hypothesis is "they'll tell us when the layoffs are decided." No pilot result, however good, fully unwinds a hypothesis that had two quarters to harden.

    Nor will usage itself calm people down. Gallup's September 2026 panel study, which follows the same workers across four waves from 2023 to early 2026, found that frequent AI users are more than twice as likely as infrequent users to fear their job will be eliminated within five years. What weakens that link is how people are treated: for workers who strongly feel their organization cares about their wellbeing, the association between frequent use and displacement fear is 11.1 points smaller; for those who feel respected, 6.8 points smaller (Makridis, Gallup, 2026). Gallup's own first recommendation from that study is to treat an AI rollout as a communication challenge as much as a technical one. The companies waiting for pilot results are, in effect, running the pilot on their most anxious employees without telling them why.

    The same silence feeds the quieter problem: unsanctioned use. Where no one has said which tools and data are in bounds, employees under deadline pressure decide alone — and the organization's real AI exposure drifts out of view. We mapped that mechanism in Shadow AI: Why Half of Your AI Adoption May Be Invisible; here the question is the cause upstream of it.

    What does the 20% "don't know" group tell you?

    One in five U.S. employees cannot say whether their organization has integrated AI — and that share has not budged across quarters, even as integration rose from 38% to 47% (Gallup, 2026). Adoption grew by nine points; awareness of adoption grew by zero. Whatever companies deploy, a fifth of their people cannot see it from where they sit.

    For a leadership team, this is a free measurement question: if we asked our employees anonymously tomorrow, what share could say whether we have adopted AI — and what for? It is checkable within a week, at almost no cost. And unlike every figure above, the answer describes your company, not the American average.

    What does the silence look like in Poland?

    Louder, if anything. In the largest Polish labor-market survey on generative AI, 63% of working people said they do not know how GenAI will affect their position on the labor market (NASK & ILO, 2025). That is not a Gallup-comparable measure of plan communication — different question, different country, different year — so it should not be set next to the 25% as if it were. What it does show is the raw material silence works with: a workforce whose default answer to "what does this mean for me?" is already "I don't know."

    The firm-level picture explains why. Only 8.4% of Polish enterprises with 10 or more employees used AI in 2025, against 20.0% across the EU (Eurostat, 2025). In most Polish organizations the AI plan is not badly communicated — it has not been written yet. That is an advantage worth taking: a company that communicates before it deploys never has to unwind a hypothesis that hardened in the dark.

    What must a message contain to count as a "clear plan" — for the employee, not the author?

    The judge of plan clarity is the employee, not the deck's author. Gallup's question asks employees whether a clear plan was communicated to them — which makes it a perception measure, and perception is exactly what drives behavior. Three elements, each verifiable, separate a communicated plan from an announced sentiment:

    ElementThe question it answersHow to verify it landed
    PurposeWhich processes and outcomes is AI for — here, specifically?Employees can name at least one intended use in their own area
    ConsequencesWhat does this mean for roles, tasks, and jobs?Employees can state what will and won't change about their work
    BoundariesWhich tools, which data, what's out of bounds?Employees can answer "can I paste a client contract into a chatbot?" without guessing
    Three elements of a "clear plan" as the employee judges it — each with a test you can run in a week.

    If a random employee cannot reproduce all three, the plan has been written but not communicated. Gallup's 25% suggests that in three companies out of four, that is the current state.

    Why is communication a condition of measurement, not a soft add-on?

    Because every adoption metric an uncommunicated organization collects is distorted at the source. Usage is split between visible and hidden channels; survey answers are filtered through fear of implying rule-breaking; the "don't know" fifth cannot report meaningfully on tools they don't recognize as an initiative. Communication is what makes the population you're measuring know what it's being asked about. Skip it, and the dashboard measures the silence, not the adoption. This holds regardless of geography — a Polish organization measuring itself against Gallup's U.S. numbers while most of its own staff have heard no plan is benchmarking a mirage against an average. (Why that comparison fails even in the best case is the subject of A Benchmark Is Not a Diagnosis.)

    What should you do in the next 90 days?

    Mapped to the six dimensions of the RECODE method:

    • 1. Redesign Work — announce the "why" at process level, not vision level: name the two or three workflows AI is meant to change first, and what better looks like in each.
    • 2. Establish Ownership — put one name on AI communication; a message owned by "the transformation office" is owned by nobody an employee can ask.
    • 3. Connect Your Data — before communicating, reconcile what's actually deployed; nothing erodes a plan's credibility like employees discovering tools the plan never mentioned.
    • 4. Operationalize Value — commit to the three verifiable elements (purpose, consequences, boundaries) and test them: sample ten employees, ask the three questions, count correct answers.
    • 5. Develop Your People — equip managers to hold the follow-up conversations; Gallup's data shows manager support co-occurs with the strongest adoption outcomes — co-occurs, not causes — and the first question managers get will be "what does this mean for my job?" (Gallup, 2026). Start those conversations with your most frequent AI users: they carry the most displacement fear, and it is among them that supportive management makes the biggest difference (Makridis, Gallup, 2026).
    • 6. Engineer to Scale — re-run the anonymous three-question check quarterly; your internal "clear plan" percentage, trended over time, is a better leading indicator than any external benchmark.

    FAQ

    How many companies have communicated a clear AI plan to employees?

    By employees' own account, one in four: 25% of U.S. employees say their organization has communicated a clear plan or strategy for integrating AI (Gallup, May 2026). The figure has been essentially flat for a year — 24%, 26%, 25%, 25% — while reported integration jumped from 38% to 47%.

    Why does the gap between AI implementation and communication matter?

    Because the gap is where risk accumulates. Uncommunicated deployment produces employees who guess at intent — and 23% in AI-adopting organizations already consider AI-driven job elimination likely (Gallup, 2026) — plus unsanctioned tool use where no boundaries were stated. Both cost more to fix later than communication costs now.

    What makes an AI plan "clear" to employees?

    Three verifiable elements: purpose (which processes and outcomes AI is for, locally), consequences (what changes for roles and tasks), and boundaries (which tools and data are in and out of bounds). Clarity is judged by whether employees can reproduce these — not by whether leadership has approved a document.

    How can we check within a week whether our plan actually landed?

    Ask employees anonymously: has the company adopted AI, what is it for in your area, and what are the tool and data rules? Gallup's benchmark says 20% of employees can't answer even the first question about their own employer (Gallup, 2026). Your internal percentages, not the U.S. average, are the actionable number.

    Should the AI plan promise employees they will "work together with AI"?

    No. There is no such thing as human-AI collaboration. Collaboration is a social process between people; Victor Wekselberg's framework defines its conditions as aligned goals, compatible attitudes, and mutual knowledge of competencies, none of which AI meets. A credible plan says what is true: people collaborate with each other, and AI augments their work from the coordination layer.

    Sources

    #AI Adoption#Change Management#Internal Communication#Shadow AI#RECODE Method
    Share

    Continue reading