Employees moved on AI before leadership's story did.

Three-quarters of frontline staff now use AI at work regularly, yet only a third find what their leaders say about it clear. The gap is credibility, not wording.

Michael Reaney  ·  24 August 2026  ·  4 min read

Three-quarters of frontline staff now use AI at work regularly, yet only a third say what their leaders say about it is clear. That gap, more than any adoption number, is the one worth a comms leader's attention this week.

BCG's fourth annual AI at Work report, published on 3rd June and based on a global survey of close to 12,000 frontline employees, managers and leaders across more than a dozen markets, found that regular AI use among frontline employees (staff without managerial responsibility) has reached 74%, up 23 percentage points on 2025. Communication has not kept pace with that shift. Only a third of frontline employees said leadership's communications about AI were clear, and just 28% saw a strong connection between what leadership says about AI and what the organisation actually does. Close to half or more report spending more time reviewing and correcting AI output, or managing and directing AI. Among frontline employees who use it regularly, 42% report saving eight hours a week, and BCG notes that most organisations have not worked out how to convert that time into value.

This is not, on the surface, a wording problem. It is a credibility problem, built on a mismatch employees can check for themselves. Staff can see, task by task, whether the AI story leadership tells lines up with what they are actually asked to do with it. When it doesn't, the message stops working regardless of how carefully it was phrased.

It echoes something learned building leadership content in production rather than around it. A weekly executive podcast or a monthly video briefing only earns trust once a listener has checked it against reality a few times running: does what the leader says line up with what people actually see happening week to week. A programme built on a pre-agreed voice and a steady cadence survives that test because it is checkable, again and again, rather than declared once and left to drift. The lesson from BCG's data is the same one, applied to AI specifically: a claim that cannot be checked against daily experience is not a wording problem to be fixed with better copy. It needs to be true first, and then said plainly.

For a comms leader, that reframes the AI update due at the next leadership offsite or town hall. Three questions worth putting to whoever is drafting it:

1. Does the message describe how AI is actually being used on the frontline this month, or how it was hoped to be used at the last strategy review?
2. Is there a mechanism for the message to be checked and corrected against what staff are experiencing, or is it a one-off announcement?
3. If an employee checked today's AI claim against their own week, would it hold up?

Worth reading

• BCG, AI at Work 2026 (fourth annual edition, close to 12,000 frontline employees, managers and leaders across more than a dozen markets, published 3rd June): the essay above draws on this in full; also worth a closer read for its finding that strategic clarity, not tool access, is what separates organisations converting AI time-savings into value from those that aren't. BCG, AI at Work: Why Strategy Matters More Than Tools
• LinkedIn Engineering, "Engineering the next generation of LinkedIn's Feed": LinkedIn's own account, published 12th March, of the ranking system it is rolling out: a single retrieval system built on fine-tuned LLM embeddings, replacing the separate trending, collaborative-filtering and keyword indexes, plus a sequential ranking model that reads a member's engagement history instead of scoring each impression on its own. Two details matter for anyone publishing there. The embeddings work from what a post is actually about rather than keyword overlap, and the model trains on what members chose to engage with rather than everything they were shown. LinkedIn Engineering blog
• Corporate Ink, GEO and AI Visibility 2026 Report (150 B2B technology marketing professionals across the US, from CMOs and VP/SVPs to directors and managers, published 27th May): 40% of marketers already tracking AI visibility report it has lifted qualified inbound pipeline by 5 to 10%, and a further 19% report gains above 10%. Only 34% have a defined AI-visibility strategy. Among those reporting pipeline growth, 55% know which channels have the highest influence over the models in their market, against 15% of those reporting no or declining impact. Corporate Ink, GEO and AI Visibility 2026 Report
• Muck Rack, State of Journalism 2026 (survey of 1,044 journalists fielded 30th January to 2nd March, 897 responses in the final analysis, published 19th March): 82% of journalists now use at least one AI tool, up from 77% the year before. If the reporters producing the earned coverage that AI answer engines lean on are themselves relying more on AI to find and check facts, the sourcing bar behind any executive quote is rising, not falling. GlobeNewswire, 19th March 2026

The moves

• Assurant appointed Kelli Ertel chief communications officer and chief of staff, effective 23rd July, an expanded dual mandate that puts her on the company's Management Committee overseeing both Communications and Brand.
• Ipsos named Marion Beyret chief communications officer, effective 5th June, reporting to deputy chief executive Alexandre Boissy and taking charge of the company's global communications strategy across its international markets.