AI summaries overtake workplace communications, Fresh says

Fresh has published research on the use of artificial intelligence in workplace communications, highlighting a widening gap between what organisations send and what employees absorb.

Based on a survey of 1,000 professionals in the UK and US, the report examines both the people producing internal communications and those receiving them. It describes what Fresh calls an Employee Attention Recession, driven by the volume of material workers are expected to process.

The findings show that 83% of employees believe they receive too much internal information, while 35% say they can no longer keep up. Only 12% now read internal communications in full, the least common response recorded in the research.

AI filter

Artificial intelligence summaries have overtaken more established channels as the main way employees discover company news, according to the study. Some 28% said they now encounter company updates via AI first, ahead of the company intranet, leadership emails or word of mouth.

This shift has created new tension between workers and communications teams. While 95% of employees said they trust AI to capture key points accurately, 92% of internal communications professionals said automated summaries distort emotional tone and intent.

Fresh also found that 88% of employees are using AI to summarise workplace communications. At the same time, 98.6% of internal communications professionals said they know their workforce is using AI to filter messages, yet only 33% of organisations have formal monitoring or policies governing how content is reinterpreted.

The findings suggest many corporate messages are being shortened, filtered or reframed by software before employees engage with the original version. That raises questions for leadership teams that rely on internal channels to communicate strategy, culture and change.

David Bowman, Product Director at Fresh, said the issue has been building inside internal communications teams for some time.

"I've spent years at internal comms events, and as a vendor, people tell you the truth," Bowman said.

"The truth is: most communicators already know which content is noise. Awareness is a problem, but so is the fact that 'no' is a high-stakes word when the person making the request is also the link between you, the information you need, and the employees who need it. This report should give practitioners the data to say it."

Volume problem

The report presents the issue less as a technical shift than as a behavioural one. Faced with a constant stream of updates, employees appear to be rationing their attention and relying on tools to condense information before deciding what to read.

For employers, the result is weaker control over message delivery. Leaders may believe a note or update has reached staff as intended, but the research suggests many workers see an AI-produced version first and may never return to the original text.

That matters particularly for internal announcements, where tone, emphasis and nuance shape how employees interpret management decisions. If summaries strip out context, the message received may differ sharply from the message sent.

The study also points to a governance gap. Although nearly all internal communications professionals surveyed said they were aware staff were using AI tools in this way, only a third said their organisations had put formal rules or oversight in place.

Bowman said communicators now need to account for AI as an intermediary rather than treat it as a marginal tool used by a small number of staff.

"Subtraction is now a strategic necessity," Bowman said.

"To be heard in 2026, organisations must design content to survive the summary. If you don't cut the noise, the 'Ghost in the Machine' will do it for you, and you won't like the result."

Fresh is part of Advania UK. Its research covered professionals in the UK and the US, surveying both people who create internal communications and employees who receive them.

The figures underline how quickly AI tools are moving from optional assistants to frontline filters for corporate information, with 28% of employees now seeing company news through AI before any original source.