Five stories, one uncomfortable thread: in communications, playing it safe is usually the riskiest move on the table. Chuck lines up a report where 88% of internal comms practitioners say overlapping crises damaged their wellbeing, a survey where half of employees call the volume of messages about right while 44% feel overwhelmed, a Golin index showing Fortune 250 CEOs shed nearly three trillion earned media impressions by going quiet, Meta logging its own employees’ keystrokes to train an AI, and a bonus story about shuttered startups selling their old Slack archives as training data. Jenni’s line on the last one covers most of them: just because you can, doesn’t mean you should.
Chuck opens with the Equilibrious Shifting Ground report, flagging ICology member Ellen Griley’s involvement, and the headline numbers land hard: 24 veteran practitioners, most director level and above with five or more years in the seat, of whom 88% say their personal wellbeing was affected by overlapping crises and 83% report stress or burnout. The report refuses to blame workload alone. It points instead at the mismatch between what organizations say they want from communications — strategic, transparent, trust-building — and what leadership actually permits: nearly three quarters aiming at a strategic model, 18% believing they arrived, and 61% with no formal change comms approach in a moment when every communication is effectively a change communication. Its STEADY framework (safety, trust, environment, agency, dialogue and you) rests on the argument Chuck puts to Jenni, that neutral, sanitized language is not safe communication but a transfer of risk from the organization’s legal exposure onto employees’ trust. Jenni buys it, and takes it further using her week of reading on psychological safety: dodging the real thing does not reduce questions, it multiplies them, and it removes the candor that makes an organization safe in the first place. Her route past a nervous legal team or CEO is to argue impact and consequence rather than principle — say this, and here is what happens next; if you are comfortable with that, fine. Chuck, a gambling man by his own admission, frames it as choosing who you gamble with, and notes the pattern is the same one the profession runs on engagement, where a disengaged workforce gets treated as the employees’ defect rather than the organization’s. Jenni’s own version of the gap is a joke that is not really a joke: hired as a communications director, they loved the idea of a communications director and did not enjoy the reality of one.
The Korbyt and Reworked survey of 1,175 full-time US employees produces a contradiction Chuck thinks every communicator should sit with: half say the volume they receive is about right, and 44% feel overwhelmed anyway, which only reconciles if people have gotten good enough at filtering that they no longer register the volume — with 89% only moderately confident they are not missing something important. The trust finding is the operationally useful one. 73% say the sender, not the channel or the format, decides whether a message is trusted, and the most trusted sender is the direct manager rather than senior leadership or a branded email, which is a problem for cascade communication as most organizations run it. Chuck, who advised on the results through ICology, admits he spent a decade disbelieving that finding on the grounds that managers were simply the only voice employees ever heard, and takes being proven wrong in good humor. Jenni holds both facts at once: her research with deskless and frontline workers always puts the manager first, and manager capability remains one of the profession’s biggest exposures, because a poor communicating manager colors how employees feel about the organization entirely. Why the gap never closes is, in her view, a refusal to fund the fix — nothing is on fire, so nobody acts, and when it does catch fire the blame lands anywhere but on management capability. On AI, 81% believe they can tell when something was machine-written and 45% question its accuracy the instant they suspect it, while 92% want AI used to reduce overload rather than generate more of it, the survey’s “reduce, don’t produce”. Chuck’s counterexample is a detector that scored passages of Mary Shelley’s Frankenstein as AI-written with 100% certainty, and Jenni wonders whether the 81% is really a trust measurement in disguise — confirmation bias about a colleague you already doubt, dressed up as detection.
Chuck warns he is about to sound like an external comms guy and then brings it home. Golin’s CEO Impact Index, behind its new guidelines for CEO communications, tracks the public activity of the top 250 Fortune 500 CEOs and finds they collectively lost close to three trillion earned media impressions against 2024, with 90% of that decline coming from stepping back from traditional media. The silence bought them nothing: it created a vacuum other people’s narratives filled. Donald Trump was the single biggest topic appearing alongside those CEOs in 2025 at 17.3% of earned coverage, ahead of AI at 9%, and 68% saw DEI, climate, immigration or political alignment surface as a meaningful factor in their public engagement. Golin’s best-performing posture is “engagement without endorsement”, framing decisions in business rather than political terms, which Chuck cheerfully calls weak — though he concedes the underlying finding that silence on social issues does not build a shield, it reallocates risk. The AI data is the part with direct internal relevance: 86% of CEOs now reference AI publicly, the framing has shifted from opportunity to workforce cuts, and leading with “we’re cutting jobs because of AI” generates real sentiment liability where reskilling and product innovation framings hold up. Jenni’s read for internal communicators is about occupied space — stop showing up on your internal channels and somebody else takes the ground — and she reaches for the trade-off she keeps returning to: saying yes to one thing is saying no to another, and the reverse is just as true, so what did those CEOs do with the attention they withdrew? Chuck’s answer is the sharpest line of the segment: imagine if that 90% pullback from media had been spent internally instead. They have not done that.
Meta’s internal program captures employee keystrokes and mouse clicks across hundreds of sites and apps — Google, LinkedIn, Slack, GitHub among them — to teach its models how to use a computer, and CNBC’s reporting notes the employees who called it dystopian did so in internal messages Meta was presumably also monitoring. Chuck’s point is that communication around the launch appears to have been minimal, which is itself a choice, and that the concerns raised inside — exposed passwords, unreleased product details, immigration status, health information, family data — describe exactly the trust erosion that happens when an organization moves faster than its employees can process. Jenni declines the psychological safety framing on a technicality, since nobody looks likely to be punished for speaking up, and objects on scientific grounds instead: the whole point was a big and unbiased data set of how people work, and the moment people know they are watched, or discover they have been, behavior changes. Big, probably. Unbiased, never — and she admits she would be the employee deliberately behaving strangely to poison the sample. Chuck reaches back to companies timing bathroom breaks under the banner of productivity and reminds listeners that Slack DMs are direct messages, not private ones, but the volume of unanswered questions is what he cannot get past, and Meta has banked no trust to spend on them. Jenni’s fishing net does the rest of the work: is this a net so fine it catches every last minnow, or one with holes big enough to let most of it through? Nobody has said, which is most of the problem.
The fifth story is the freq-out Chuck held over from last week because it deserved a longer conversation. Fast Company reported on Asset Hub, a product from SimpleClosure — a startup whose business is helping other startups wind down — that lets shuttered companies sell years of Slack conversations, email threads and meeting notes as AI training material. Around 100 deals have gone through, payouts run from tens of thousands into six figures, and one former founder sold what she described as a decade of internal records for hundreds of thousands of dollars. The people whose messages fill those archives were never asked, and whether a company owns and can sell its employees’ internal communications is largely untested law. SimpleClosure says the data is anonymized first; researchers and lawyers are not convinced that covers messages touching health, personal conflict or sensitive business decisions, and what a buyer is actually acquiring — an unfiltered view of how a real organization operates — is worth more than most public internet text. Chuck’s broader point is that this is not a startup story: any organization on Slack or Teams is sitting on an asset its employees have never thought of as one, and the obligation to say so has no agreed trigger. Jenni’s discomfort is about ownership and about motive — she cannot construct a version where buying another company’s internal chatter genuinely helps the buyer, lands on “just because you can doesn’t mean you should” and credits Jurassic Park, and names greed as the part that bothers her most. Chuck’s read is simply that everything has a price and a winding-down company will find value wherever it exists. Both want to hear from listeners whose employer has a policy on this, and Chuck suggests the chief legal officer is the right person to ask, since a company can own something and still choose not to sell it.
What did the Shifting Ground report find about internal comms burnout? Equilibrious Communications surveyed 24 veteran internal comms practitioners, most at director level or above with five or more years in the role. 88% said overlapping crises had affected their personal wellbeing and 83% reported stress or burnout. The report attributes it to a structural mismatch rather than workload alone: nearly three quarters aim for a strategic comms model, only 18% believe they have reached it, and 61% have no formal change comms approach.
Is neutral, sanitized language actually safer for an organization? The report’s argument, which both hosts accept, is no. Neutral language does not remove risk, it relocates it — off the organization’s legal exposure and onto employees’ trust. Jenni links it to psychological safety: vague communication generates more questions, not fewer, and removes the candor that makes an organization healthy. Her advice for persuading legal or a CEO is to argue in terms of impact and consequence rather than principle.
Who do employees trust most as a source of workplace communication? Their direct manager. In the Korbyt and Reworked survey of 1,175 US employees, 73% said the sender was the most important factor in whether they trust a message — more than the channel or the format — and the direct manager outranked senior leadership and company-branded email. Jenni’s caution is that manager capability remains one of the least funded risks in the function, so the most trusted channel is often the least supported one.
What does “reduce, don’t produce” mean for AI in internal communication? It is the survey’s shorthand for what employees actually want from AI: 92% want it used to cut information overload rather than generate more content. The same data found 81% believe they can tell when something was written by AI and 45% question its accuracy the moment they suspect it, and that undisclosed AI drafting on a sensitive topic collapses trust quickly. Chuck’s caveat is that AI detection is unreliable — one tool flagged Mary Shelley’s Frankenstein as AI-written with 100% certainty.
What did Golin’s CEO Impact Index find about CEO silence? The top 250 Fortune 500 CEOs lost close to three trillion earned media impressions in 2025 compared with 2024, roughly 90% of it from pulling back from traditional media. Golin’s conclusion is that silence on social issues does not build a shield, it reallocates risk — the vacuum simply gets filled by other people’s narratives. Donald Trump was the biggest topic appearing alongside those CEOs at 17.3% of coverage, ahead of AI at 9%.
What is Meta’s Model Capability Initiative and why is it controversial? It is an internal Meta program that captures employee keystrokes and mouse clicks across hundreds of sites and apps, including Google, LinkedIn, Slack and GitHub, to teach Meta’s AI models how to operate a computer. Employees called it dystopian in internal messages that were themselves being monitored, and raised concerns about passwords, immigration status, health and family data. Jenni’s objection is methodological: once people know they are being tracked their behavior changes, so the “big and unbiased” data set the program was designed to collect was never achievable.
Can a company sell its employees’ old Slack messages to AI companies? Legally it is largely untested, which is Jenni’s central complaint. SimpleClosure’s Asset Hub has processed around 100 deals letting shuttered companies sell internal Slack conversations, email threads and meeting notes as AI training data, with payouts from tens of thousands into six figures. Employees are not asked, and while the seller says data is anonymized first, researchers and lawyers are not convinced that covers messages touching health or personal matters.
Jenni: Welcome to Frequency, I’m Jenni Field.
Chuck: And I’m Chuck Gose. Frequency is your go-to for real talk about comms, culture, and employee experience. Beyond the buzzwords and straight to what matters. Jenni, this week we’re going to be talking about how internal comms is absorbing a crisis, the state of workplace communication, the new normal for CEO communications, and how Meta is spying on employees. And I threw in a fifth one to shake it up here, Jenni. What was going to be my original freq-out from last week I’m bringing in at the end. So please stick around for that.
Jenni: I love it, I love it. I am out and about today. So I’m at the Institute of Internal Comms Festival in the UK, which is a two day event. I popped in yesterday and I’ve been here today because I’ve been talking about the credibility gap and how comms people can close it. Helpfully, it is a 15 minute drive from my house this year, which is why I’m here, if I’m honest. It’s so close. I was like, I have to come. I thought it was worth sharing with you a couple of things from the session today. I think we’re going to probably dig into this in a few episodes’ time. But there’s a chap talking from Microsoft about the Work Trend Index, which has just come out. And he was talking about the number one success factor in AI adoption around the world being organisational culture, which I thought was an interesting piece of insight, so we’ll dig into that. But it’s been an interesting two days, lots of talk about change, lots of talk about trust, credibility and AI. They’re probably the four big topics, but we’ll talk probably a bit more about IoIC in a few weeks because their IC Index is coming out and I want to dig into that when we get a little sneaky advance copy. So go on then, kick us off with one of your five articles for today.
Chuck: Well, before we get into that, I’m just curious — if you’re Jenni Field, do you just get to pop in to places? Is that how you roll over there?
Jenni: So to be fair, they did say that to come to the whole conference you have to pay to come on the day you’re not speaking, but I said I can only come for the first half an hour, because I was speaking somewhere else yesterday at a leadership conference for a corporate client. So I snuck in for free. And I’m speaking today, so I get to be here today anyway. So yeah, just pop in. Channel Jenni Field and just pop in.
Chuck: All good, all good. Let’s kick this off with the internal comms crisis. ICology member Ellen Griley is connected to this one, so I just wanted to flag that up and share some pride there that Ellen’s working on this. The Equilibrious Shifting Ground report surveyed 24 veteran internal comms practitioners, most of them with five plus years and director level and above experience. What she found is that 88% said their personal wellbeing was affected by overlapping crises and 83% reported stress or burnout.
What’s notable is where they pin the blame. Workload is involved, but more importantly it’s the structural mismatch between what organizations say they want from comms — things like strategic, transparent, trust building — and what leadership actually allows. Nearly three quarters claim to be aiming for a strategic comms model, but only 18% believe they’ve gotten there, and 61% have no formal change comms approach at all.
In a moment where, as the report puts it, every communication is effectively a change communication, that’s a pretty big gap. The report offers a framework called STEADY with six principles: safety, trust, environment, agency, dialogue, and you. The core argument is that neutral, sanitized language isn’t safe communication, it just transfers risk from the organization’s legal exposure onto employees’ trust.
Jenni, neutrality is named as the enemy here — the idea that the sanitized, perhaps even boring language is somehow still responsible communication. Do you buy this? And if you do, how do you make the case to a legal team or CEO who’s convinced that saying less is actually safer?
Jenni: So I agree that if we’re saying that that sort of neutral, boring language isn’t safe, then I agree with that, because I think that that is too risky. Because when you’re doing that, you’re avoiding the real stuff. So all you’re doing is just asking more questions. People are just going to have more questions. It doesn’t really help anybody. I love the point about how it shifts the risk from the organisation’s legal exposure onto employees’ trust. That for me is the big thing, because if you’re trying to be safe, really what you’re doing is not being safe. I’ve been doing a lot of reading around psychological safety this week for some client stuff, and it basically says this: if you’re trying to do all of this, then you’re not having the candour, you’re not doing the stuff that actually creates a psychologically safe organisation, because you’re kind of hiding away from that. So yes, I think that’s the case.
The challenge around how do you communicate that with legal teams and a CEO who want to say less is all about the impact they want to have. That’s always the conversation for me. If you say this, this is what’s going to happen. And if you’re okay with that impact and consequence, that’s fine — maybe I don’t keep arguing, or maybe I do. But you’ve got to have that conversation about impact and consequence. Without that, I just think you’re going to have a very tricky time trying to convince them.
Chuck: Yeah, if I was a gambling man — and I am — this is the way I look at it. Anytime there’s risky communication, you’re gambling somewhere. And if you’re playing it safe, you’re then gambling with the employees and the trust. Whereas if you come out strong, if you come out with the real information, the risky stuff, then you’re taking on that responsibility. So again, this just shows some of that weakness in neutrality, of shifting that responsibility on to employees. No different than I would say how we view engagement. We put that on the shoulders of employees — there’s something wrong with them if they’re not engaged, and not something wrong with the organization. So I think this is just another example of shifting that responsibility.
Jenni: The bit that stands out for me in the report from Ellen is around what organisations say they want from communications and then what actually is allowed. And I’ve experienced this. I always joke that when I was hired as a communications director, they liked the idea of a communications director. They did not enjoy the reality of a communications director. And I think that’s so often the truth. I think people have this vision of what that’s going to be, but actually you’re asking the hard question, you’re challenging what that really means, you’re getting them to think maybe more ethically than they have before. And I think all of that is uncomfortable. And I think that that gap is where I want to see people like the IoIC, who have talked this week about advocating for internal comms — like that’s the job. That’s the bit we need to be doing more of to make sure that that’s really happening, because that gap is where our biggest risk is, I think.
Chuck: When you talked about the idea of a communicator, it reminded me of my wife, Kristen, talking about men as a concept. Sounds really good, but in practice, maybe not so much. Enough of that, enough of that. But if you want to check out this STEADY framework, we’ve got a link in the show notes, you can check out Ellen’s work on that.
We’re going to move along now to Korbyt and Reworked, who surveyed 1,175 full-time US employees and found the kind of contradiction that should make every communicator a little bit uncomfortable. Half of employees say the volume of communications they’re getting is quote unquote about right, but also 44% feel overwhelmed. You would think those two things can’t both be true, unless what’s happening is that employees have gotten so good at filtering out messages that they’re no longer registering the volume, which is not a win. 89% say they’re only moderately confident that they’re not missing important updates.
The trust data is also worth going into. 73% of workers say who a message comes from is the most important factor in deciding whether to trust it. Not the channel, not the format, but the sender. And the most trusted sender is the direct manager, not senior leadership, not a company branded email. That has real implications for how we think about cascade communication and why it keeps falling short.
And then moving on to AI: 81% of employees say they can tell when something was written by AI, and 45% question the accuracy the moment they suspect it, with 92% wanting AI to reduce information overload, not generate more content and noise. The survey’s phrase for it is quote unquote reduce, don’t produce, which is a clean shorthand for a governance conversation most organizations have not had yet. Also buried in the data: if employees found out an AI-drafted message on a sensitive topic wasn’t disclosed as such, trust collapses immediately.
Jenni, the data says that employees trust their direct manager more than any senior leader or official channel, but other research consistently shows manager capability as one of the biggest risks in the function, with most organizations providing almost no support. Why are we still not closing that gap?
Jenni: So I love all of this. I love the challenge that about half say that they’re getting about right and then another half feel overwhelmed. It’s just the joy of data — you can make it say whatever you like. I think that managers have always been important. This report saying that they trust the direct manager more than any other senior leader, and other research showing that manager capability is one of the issues — the two can exist at the same time, however paradoxical that might feel. So in all of the research I’ve done with deskless workers, frontline workers, it’s always, I trust my manager the most. If that manager is a poor communicator, that has a direct impact on how I feel about communication in the organisation as a whole, and how I feel about the organisation kind of more generally. So I can trust my manager because I don’t know maybe how good they could be, or how different that could be.
But it’s important for us to make that better as a skill set, manager capability, because that does have a direct impact. I don’t think that gap is closing, because I don’t think people are willing to invest the time and effort needed to upskill managers. I just don’t think they’re going to do it. I think everyone feels like it’s fine and nothing’s on fire, so we’ll just carry on. And then it will go on fire, and then they’ll blame something else for that fire. So it will never come back to being the managers, it will always be something else that’s the problem, because that’s hard and messy and costly and nobody wants to do that.
Chuck: I should have mentioned this too. Through ICology, I got to advise them on the results of this and start to dissect it and look at some of the trends, look at some of the figures. And one of the things that — it happens rarely, Jenni, but I do love being proven wrong from time to time. It’s not often. That’s why I enjoy it. I remember a decade ago or however long ago when people would say, my manager is my most trusted person. I never believed that. And I think I said, I think they say that because that’s the only person they actually hear from is their manager. So who else are they going to trust? But I think this just shows that that is such a key channel. So how we empower and equip managers with information and skills is absolutely critical.
And the other part that I find a bit comical and troubling is this, that 81% of employees say they can tell when something was written by AI. They think they can tell when something was written by AI, because I saw this video where this guy put paragraphs of Mary Shelley’s Frankenstein into an AI tool and it said with 100% certainty it was written by AI. Which, I don’t know if there’s time travel involved or what, but I’m guessing that we just don’t know all the time. We really don’t know. But it’s funny that now there’s this subjection that is happening — if we think it is AI, then we question the accuracy, but we don’t question the accuracy if it’s a person doing it, when we should question the accuracy at all times.
Jenni: But I wonder if it’s linked to a bigger trust problem. If 81% of people say they can tell when something’s written by AI, are those same 81% of people people that don’t trust the person that that’s come from? Is it just you’re having that confirmation bias? You’re going to assume that it’s AI because it’s that person and you don’t trust them anyway, and they’re lazy, and you know they wouldn’t have — do you know what I mean? You can create that story, I think, quite easily. And I wonder if that’s what’s happening.
And in fact, this reminds me of one of my thoughts yesterday when I was sat listening to the IC Index sort of top level reports. And I thought it’s really hard when you look at individual reports in isolation, because we talk about so many reports on Frequency. Every time there was something coming through yesterday, I was like, yes, but — yes, but in terms of other reference points and other things. You really have to look at all of it to tell the whole story. Otherwise you’re getting such a narrow view of the reality, because there’s so many different parts playing a role. And I think that’s important when we’re looking at data like this: there’s other things that play that will be linking to some of those numbers.
Chuck: There’s also a lot of really interesting hybrid data in here. I know we’ve talked a lot about RTO in the past. I didn’t want to bring that in, but if you want to learn more about this report, we’ve got it linked and you can check out that hybrid data as well.
On to this next one. It’s going to sound like I’m talking external comms at first, and I am, but don’t worry, I’ll bring it back. Golin tracks the public communications activity of the top 250 Fortune 500 CEOs through their CEO Impact Index, and the 2025 numbers are interesting. Those 250 CEOs collectively lost nearly three trillion earned media impressions compared to 2024. 90% of that came from pulling back from traditional media. The silence didn’t protect them, it just created a vacuum that filled with other narratives. And for those that are in internal comms, this is a version of a story we already know.
The political dimension is hard to ignore. Donald Trump was the single biggest topic appearing alongside Fortune 250 CEOs in 2025, cited in 17.3% of earned coverage, which beat AI at 9%. 68% of CEOs had DEI, climate, immigration and political alignment surface as a meaningful factor in their public engagement. The approach that held up best, per Golin, was what they call quote unquote engagement without endorsement — framing decisions in business terms rather than political ones. That also sounds weak AF. Their read: CEO silence on social issues, quote, doesn’t build a shield, it reallocates risk. There’s that risk word again.
The AI finding is the one with the most direct IC relevance. 86% of CEOs now reference AI in public communication, but the news has shifted from opportunity to workforce cuts, which we’ve talked about on here. Golin is blunt: leading with quote unquote we’re cutting jobs because of AI generates real sentiment liability. Organizations that anchored that message in reskilling or product innovation fared considerably better. This is not just a PR lesson. It applies to how internal comms teams are also handling AI narratives internally. Jenni, what else can internal comms learn from this Golin data?
Jenni: So it’s interesting, isn’t it? Because I think it’s really telling that story about silence, and it’s telling the story about the importance of how you message something — maybe thinking you’re mitigating risk, but there’s unintended consequences. And I think we see that so much in internal comms. The AI stuff we talked about, I think either last week or the week before, about how you’re framing that has a direct impact on adoption and stuff. And I think this is kind of referencing the same.
I think the silence stuff is interesting. I think the silence piece around the fact that it’s just created other people to step into that space. And from an internal comms perspective, you’ve got to think about that. If you’re not speaking on some of those digital channels that you might have — I’m not going to name any, but if you think about all of the various different ways and platforms that exist where you can communicate internally, if you stop doing that, or you’re not visible there as a leader, someone else is going to fill that space. And I think that’s something to reflect on in terms of the impact that has here as well. So there’s definitely something about risk, silence and framing that I think this is relevant for internal comms.
Chuck: I think this goes back also to that neutrality thing that we talked about earlier, whereas if you’re safe, that’s not a shield. You’re just putting that risk somewhere else. And I think that shows the weakness of some leaders willing to put that onto others instead of themselves. We talked about that with what was happening in Minneapolis with ICE, and the kind of weak message of some of those leaders that came out. I think that’s what people want from their CEO, COO, whoever it is that’s out there speaking.
I think the other part that was interesting to me was the fact that Donald Trump mentions nearly doubled AI. Which, if you’re saying you want to play it safe, I wouldn’t mention Donald Trump in your communication. So it’s really interesting to see that this is the top of the top CEOs, not from a performance level, but from a large organization. I think this just shows that pulling back from traditional media, like you said, that’s creating a vacuum. That’s going to be filled with something else.
Jenni: Yeah, and I think we always have to remember that sort of mantra or phrase which I heard a long time ago, and I don’t know where it started from: if you’re saying yes to one thing, you’re saying no to something else. And I think sometimes we forget that. We’re very good at saying yes and yes and yes and doing more and more without thinking about what you’re then saying no to. But equally, if you’re going to say no to stuff, then what are you doing instead? So I don’t know whether with the silence it meant people were going off doing something else and that had a bigger impact. So it’s that trade-off that I think we have to be mindful of.
Chuck: Imagine if that 90% pulling back from traditional media, they then spent that time internally. What a difference that would make inside organizations.
Jenni: Yes, they haven’t done that. I’m pretty sure that they have not done that. Not sure that’s where they went.
Chuck: They have not done that. Speaking of interesting leadership, Meta launched something internally called the Model Capability Initiative. And when employees found out about what it involved, some of them called it dystopian — in internal messages that Meta was presumably also tracking. The program captures employee keystrokes and mouse clicks across hundreds of sites and apps, including Google, LinkedIn, Slack, and GitHub, with the stated goal of teaching its AI models how to use computers. To do that well, you need a quote unquote big and unbiased data set of how people work. But now the employees are the data set. Their communications around this launch appear to have been minimal, which is also a choice. Speaking of silence.
Internal concerns raised by employees include the potential exposure of passwords, unreleased product details, immigration status, health information, and even family data. The tracked site list reportedly included ChatGPT and Claude before both were removed. Whether you find the surveillance itself acceptable or not, the way it was communicated — or wasn’t communicated — is a case study in what trust erosion looks like when you move faster than your employees can process. For organizations watching how this AI employee relationship gets defined right now, this is 100% worth paying attention to.
Jenni, for years we’ve heard stories about companies monitoring employees. The employees who called this dystopian were doing so in internal messages that Meta was also monitoring. What does that tell us about the state of psychological safety inside organizations that are simultaneously asking for trust?
Jenni: So first of all, psychological safety is about being able to speak up, speak freely, have candour without fear of punishment. I don’t think that’s the issue here, because I don’t think people are going to get punished for anything. So I don’t think that’s the challenge. The issue I have with this is more scientific, from a research perspective. The reason for this is that they wanted to educate AI. So they were like, the goal is to teach AI how to use computers. So they needed — to use the quotes in this — a big and unbiased data set, right? Except it is not, right? So even if you had communicated that it was going to happen, it leaked out anyway. But if you tell somebody we’re going to start tracking you, you’re going to start behaving differently. It’s a natural human reaction. If you then find out that you have been being tracked, you’re going to behave differently. So nothing about this is unbiased. It’s probably big, but this is not actually going to do anything.
And if I found out this, I would start doing things really oddly in the hope that it sabotages stuff, because I’m that person. But I think that’s the biggest issue for me: I don’t know that the goal and the purpose of this was right even in the first place, in terms of the communication, in terms of the ethics around it. All of that is questionable. But is that really the goal, the big unbiased data? I’m not sure you were ever going to get that, and I’m not sure that’s the right goal, and I’ve got bigger questions about that and scientifically how you’re doing it.
Chuck: I mean, I remember years ago hearing stories about companies allegedly tracking employee bathroom breaks and all this kind of stuff, and they labeled it under the thing of productivity. So I’m not surprised by any of this. I think it’s a good reminder to employees out there — for Slack, for example, they’re called Slack DMs, not PMs. They’re direct messages, not private messages. At any point in time your company can go in and see what you’re typing, what you’re doing, what those messages are. So this shouldn’t come as a surprise. I just think when you think about the types of relationships that people have, they might be sharing sensitive information. We’ve said health information, immigration status, is it passwords? All of those things that you might be logging into — your personal bank account — do they now have that information stored? There’s a lot of questions that I think Meta hasn’t done anything to earn anyone’s trust on, let alone their own employees’. And so that’s where I think that the safety part stands out. I don’t know that your employees would feel safe, psychologically safe, at a workplace like that.
Jenni: You’ve got no trust credits in the bank, have you? It’s not like someone’s gone, this is absolutely fine, because we’ve spent years building trust in this relationship and I know what you’re going to do, because I understand what we’re doing and why we’re doing it. Everyone’s just gone, what are we doing? Because it’s come from a culture and a leadership style and an approach that doesn’t value employees, doesn’t seem to have that ethical grounding and all of those things. So it’s interesting that the programme captures these things. What does it capture? I’m thinking of a net when you go fishing — is it a net that’s got hardly any hole, so you get all the fish, even the tiny tiny fish? Or is it one that’s just got big holes, so you just get the big fish and all the little ones can go away? What are we capturing? I think there’s too many questions, and I think it is just overall just a very bad idea and not particularly thought through.
Chuck: All right, we’re going to move along to our bonus content this week, Jenni. Number five. This was going to be my freq-out last week, but I wanted us to have a longer conversation about it. In a previous episode, we talked about Anthropic scanning old books to train its model. But this is something different — or is it? I can’t quite determine the difference between the two.
Fast Company reported on a product called Asset Hub, built by a company called SimpleClosure, a startup that helps other startups wind down. There’s an interesting concept: I’m going to be a startup to help other startups close. Anyway, there’s a business for everything. This business, though — they help shuttered companies sell their internal data as AI training material. This means years of those Slack conversations, email threads and meeting notes can be sold to AI companies looking for authentic organizational data to train on. About 100 deals have been processed so far. Payouts range from tens of thousands to well into six figures. One former founder sold what she described as a decade of internal records for hundreds of thousands of dollars.
The employees whose messages are in those archives were not asked. The legal question of whether a company owns its employees’ internal communications and can sell them is largely untested. SimpleClosure says that data gets anonymized before sale. Researchers and lawyers aren’t fully convinced that covers it, especially for messages that touch on things we just talked about — health issues, personal information, sensitive business decisions. What buyers are getting is an unfiltered view of how real organizations operate. That’s worth more to an AI company than most public internet data.
The bigger question isn’t about startups. Any organization running Slack or Teams now is sitting on a data asset its employees have never thought of as one. If the company gets acquired, merges or shuts down, who owns that archive? If it already has real monetary value, which apparently it does, when does the obligation to tell employees kick in? Jenni, are you surprised by this? We work for ourselves now, but we’ve been employees before. How do you feel about this happening?
Jenni: So this whole thing makes me feel really uncomfortable. And it makes me feel uncomfortable because it’s the question of, can you sell employees’ internal communications? Who owns that communication? Who owns the email? That’s the big question for me. I don’t think it’s ever been seen as data that’s helpful before in that way, and I think that’s a part of AI that is a big question. I agree that the data is anonymised before sale — and therefore it’s fine? I’m not sure I’m here for that.
But my biggest question is why? I get it if you’re trying to train AI stuff or understand stuff, but if I was a startup getting an insight into data of another company, I’m not sure why I would do that, I’m not sure why that’s helpful or what that’s helping me achieve as an organisation. Just because you can doesn’t mean you should. I stand by Jurassic Park’s ethos forever. So there’s something about why. I can’t figure out why this is truly helpful for organisations. And I really don’t like it. And I would like someone legally to figure out who owns the employee’s internal communication and that selling of it. I think that’s a massive legal question.
Chuck: I think this shows that everyone and everything has a price, and as startups are winding down and they’re selling these things off for parts, it’s clear there’s value there. Whether it’s training ChatGPT or Claude or Gemini or whatever’s buying it, maybe it’s some other proprietary AI tool — whoever’s buying that is getting data that they cannot get anywhere else. So that’s where that value comes from.
Jenni: It’s also just the ethics. When you were saying about there was one former founder who sold what she described as a decade of internal records for hundreds of thousands of dollars — I feel like that’s not yours to sell. I just feel like there’s this greed element to this, which is what probably makes me most uncomfortable, because it probably goes to one of my values.
Chuck: Jenni, wait a minute — are you saying people are motivated by money? What?
Jenni: And I think that’s what’s making me uncomfortable. This just feels greedy. It feels like a greedy thing, and I don’t know how that’s helping us move society forward. Maybe I’ve just got grand ideas of what we should do, but it feels dirty and greedy and I don’t like it.
Chuck: I would love to hear from people if their company has a policy on this or has even talked about this. And if not, maybe this is a fantastic conversation to have with your chief legal officer, to truly understand — it could be that the company owns it, but the company doesn’t sell everything that it owns. But it shows that maybe in certain cases, everything is for sale.
Jenni: Yeah, I don’t like it. But if this is happening to you, then let us know, because we would like to hear more about it. And maybe you’re listening to this and thinking, I don’t know what you’re getting so worried about, Jenni, it’s absolutely fine, calm down. Also let me know, because it might just be one of those days today.
Chuck: But this is why I was freq-ing out about it. One, because I never thought of it before. Then two, the audacity. But then the realization of, yeah, when these companies wind down, they’re looking to pay off all debt, they’re looking for value anywhere, they’ll get it. And the fact that still, the irony of a startup helping other startups close. That’s just —
Jenni: Yeah, blows my mind. But you raise a question in there around, if we think about 10 years ago, companies closing down, what happened to all that data? Where is that? Maybe they’re all going to go back and go, hang on a minute, I’ve still got that on a floppy disk.
Chuck: They’re probably now wondering why didn’t we sell it. Where is that server that somebody had? Yeah, we don’t need to give anybody any ideas. That wraps up our content for this week, Jenni. Thanks for letting me throw that last one in there. Let’s move on to our freq-outs. What are you freq-ing out about this week?
Jenni: So my freq-out this week is a little bit relevant, because I’ve been speaking a lot this week, and it’s about speaker vetting. And I say this because I was chatting to one of the speakers at the conference this week who heads up internal comms in an organisation, and she said to me, you know, I just put my hat in the ring and they just said yes. Nobody knows if I’m good or not. I don’t know if I’m good or not, because I’ve not done this before. And she’s like, and they just said yes. So we had a chat about that, and it just got me thinking about events in general, how speakers are picked. I don’t think anywhere has got that right. IABC — we’ve talked about that before, I definitely don’t think they’ve got that right. But I wonder how you think we do that to make sure it’s a really great experience for people that are paying to come to events. It’s just left me with a bit of a question about how to do it well.
And this was someone who is a client, so we’d had a chat about speaking beforehand, because she was like, could you give me some tips? So we had a chat about it, and I messaged her yesterday and said, how did it go? And she said it was really good and she’s up for doing it again. So that’s really great. But I think her questions are valid around, well, I just put my hat in the ring and they said yes. Which just made me laugh.
Chuck: We’ve talked about that before, where you see people that are speaking and it’s who they work for that is the driver, it’s the logo. Versus people who are great speakers but maybe where they work is not a name brand, and so they have to work harder to get there. I’ve seen that all the time. I fixed this with Flyover Festival, Jenni, because I quote unquote vet the speakers, because these are people that I know. I don’t put out public asks. I don’t do anything like that. I go and hand pick the people that I know, that I’ve seen, that I know will deliver that experience.
Jenni: Which I think works when you are networked well enough to bring diverse voices in, and you’re really good at that. And I don’t think everybody makes the time to do that, and that’s why it ends up being a public call, because you want to make sure you do get the diverse voices and that’s the easiest way to do it. But yes, I don’t know how we solve that from a mass scale perspective, but it just made me chuckle. So that was my freq-out this week. What’s yours?
Chuck: Mine is, I’ve been thinking a lot about internships recently, because our daughter is home from college and she starts her — I will say first internship. She had one last summer, but this is the one that she got on her own. She went out, saw the opening, applied, interviewed, got the internship. So she is doing that. I was reflecting back on — I did four different internships when I was in college, and just what a great experience they all were in one way or another. One of them was, I went to college thinking I was going to be a newspaper reporter, did an internship at a business newspaper, hated it, and realized, okay, this is not for me. But the internship taught me that. And I’m just excited to see what she now learns in these early bedrock moments as she builds her career.
Some of the people I still keep in touch with that I interned with — this is now 30 years ago. This is pre-LinkedIn. This is before you could track people down. And we still have email addresses. Some are Yahoo. Some are some pretty obscure ones, but they’re still out there. So whether you think back on your internships or you have an opportunity to provide internships to people, just what an amazing opportunity that is. I know there’s controversy around whether they’re paid or unpaid. I did both. There’s merits. Everybody should be compensated for work, so I think that’s now largely the case. But yeah, it’s just freq-ing out about my experiences and what’s going to be rolling out for our daughter this summer.
Jenni: Yeah, I think it’s nice to have the ability to test something in a sort of finite time frame like you did, and went, actually, this isn’t for me. I think that’s got to be worth doing. And I think even all of my jobs have been along the way of working in clothes shops, of working in bars, of doing an internship at a PR firm to figure out whether or not you want to do it. I think it’s a good thing. Like you said, compensated now, but I think it’s a great thing. I’m excited for her.
Chuck: Yeah, I was thinking back on the newspaper experience. It wasn’t the work — the work I enjoyed doing. It was seeing how everybody else hated their life. They just did not enjoy their life. And so I’m like, okay, this is not going to be good for me then either.
Well, that’s what we are freq-ing out about this week. The show notes have all the articles and links from today’s conversation. If Frequency is a regular part of your week, a quick review for us goes a long way. Subscribe wherever you’re listening and share this with one person who’s interested in what we talked about today. Thank you to Poet Ali for the music. We’re back every Monday. See you next week.
Chuck is a US-based internal communications strategist and the founder of ICology – a community and resource hub for IC professionals. He brings a practitioner lens to every conversation. Chuck is a recognised voice in the industry, a regular speaker and event host, and one of the most connected people in the North American IC world.
Jenni is a UK-based leadership and internal communications consultant, author of two best-selling books, and international speaker. She runs Redefining Communications, a consultancy working with organisations around the world to help them communicate better and close the gap between what leaders say and what employees experience.
