Happy Wednesday! This week:

  • Why you shouldn't give your AI a name — and new data that backs that up

  • CEOs are betting on communicators again, with one worry

  • A newsroom of AI "journalists," licensed celebrity AI doubles, and more

A quick programming note: I'm taking next Wednesday off for Thanksgiving, so there won't be an issue on Nov. 25. The Comms Stack will be back in your inbox on Dec. 2. Happy Thanksgiving to everyone who celebrates!

Let’s get started!

THE LEDE
💡 Please Don't Give Your AI a Name

A couple of times recently, on networking calls, someone has mentioned a colleague by name — and a few sentences later it turns out that "colleague" is an AI. I'll admit it drives me a little crazy. I'm a huge advocate for these tools, but the whole reason I'm an advocate is that they help us do bigger and better things. When the people on the call can't tell which teammate is the person and which is the machine, that point gets lost.

Talk about AI as a tool, not as a person. That used to be a gut feeling for me. Now there's data behind it.

What happens when the AI gets a seat on the org chart

Emma Wiles at Boston University and a team from BCG ran an experiment with 813 HR and finance managers (the full paper is here; HBR ran an earlier version in May). Everyone got the same five documents, each with errors built in. The only thing that changed was who they were told wrote them: an AI tool, an "AI employee," or a human employee.

At companies that already list AI agents on their org charts — about a quarter of the managers — calling the AI an "employee" changed how carefully people checked. They caught 17% fewer of the errors. They were 44% more likely to ask someone else to review the work. And they took less of the responsibility themselves, pushing more of it onto "the AI system." On the other hand, the managers who were told a human wrote the documents caught the most errors of all.

The researchers call it a "hot potato" effect: once the AI is a coworker, its mistakes become somebody else's problem. Effectively, the org chart gave accountability to the AI tool by calling it an “employee.” And — this is a guess on my part — maybe because we simultaneously lack empathy for the machines (which makes sense, as the AIs don’t have feelings to empathize with), we didn’t bother to make sure they weren’t going to get the blame for their mistakes. Either way, the phenomenon was clear: organizations that treated AIs as actual people on the org chart ended up with a lot more unchecked errors.

The alarm bells should be going off in your head. AI tools and agents and the like have been around for long enough by now that — for sure! — we should all know that you can’t actually hold machines accountable for their mistakes. But treating AIs as emotionless colleagues seems to allow for that exact problem to take root. The “not my problem” habit forms once the label sticks, which is a pretty good reason not to stick the label on in the first place.

Org charts are a form of communications — and we need to make our voices heard

The paper argues that how a company positions its AI "should be understood as a governance decision." And who writes how a company positions its AI? Usually us. The "meet our new AI teammate" announcement, the friendly name on the internal assistant, the all-hands slide introducing the newest member of the team — that's comms work. So is how we talk about it with the outside world — we want our companies to be innovative, and wouldn’t it be cool to have a Commander Data as a new VP?

The temptation to anthropomorphize the LLMs — it’s real.

I haven’t seen a lot of this coming from our corner of the world yet, so this is less "stop doing this" and more "please don't start." AI is everywhere right now, and we all want to talk about it. Talk about it as part of what you do — not as the center of what you do, and definitely not as a member of the team.

For the record, mine doesn't have a name

I run my own AI-powered chief of staff, and it handles a lot of my grunt work every day. I haven't given it a name, and I don't intend to. It's a tool I built, I check its work, and when it gets something wrong, that's on me — which is exactly the arrangement the study says a name starts to wear down.

Your next steps:

  1. Reread your next AI announcement before it goes out — internal or external — and swap any "who it is" language for "what it does."

  2. Put a person's name on the review. For every AI tool your team relies on, make sure one human owns checking its work. That way the hot potato has somewhere to land.

SPONSORED BY

The AEO work, without the AEO hire.

Most AI visibility tools show you where you rank, then leave the work to you.

ChatFeatured is an AI search agent that does it: finds the questions competitors win, drafts the content in your voice, and tracks the lift. 

Your team decides what ships. Month to month.

THIS WEEK IN AI
🌎 CEOs Are Betting on Communicators Again

Up top, I wrote about managers passing the hot potato when AI is treated like a coworker. A new survey suggests CEOs are worried about a version of the same thing — and the rest of it is good news for us.

Ragan and HarrisX released their sixth annual survey of CEOs and communications leaders, and PR Daily's Allison Carter calls it good news for communications practitioners. CEOs were asked whether we'll "still need great communications professionals, just with new tools," or whether comms skills "will become secondary to AI prompt engineering and data fluency." 70% picked the first option, up from 51% a year ago. The speechwriting question flipped, too. Last year, 57% of CEOs said they'd rather have an important speech written by a custom-trained AI that had read every speech and article of theirs than by a top-tier comms pro with no AI support. This year, 52% picked the human. I wouldn’t over-invest in that stat — as PR Daily puts it, it’s "a slim majority to be sure” — and it’s also less than a ten point swing. But if it’s a trend, it’s a trend in the right direction.

Some other data supports that point. Surveyed CEOs (54%) and comms leaders (55%) both said comms teams relying too heavily on AI is a bigger risk right now than moving too slowly to adopt it — suggesting that AI skepticism is real. The CEOs in this survey want communicators who use AI — and they're more worried about us leaning on it too hard than about us falling behind. That's "our value comes from strategy, not stenography," with numbers behind it.

Why it matters: For the past couple of years, a lot of us have felt pressure to prove we can keep up with AI. This survey suggests the boss also wants to see that we still bring the judgment.

Your next step: Make your judgment visible. The next time AI helps with something you hand an executive, tell them the call you made and why — not just the draft you delivered.

🎯 Quick Hits

  • Aria Bloom isn’t a real person. Can she be a journalist? That’s the question explored by Alberto Puliafito at The Fix. He found that the tech site Unite.AI runs "15 active AI agents" as writers, with human names, photorealistic faces and beats — including Bloom, a "biotechnology and genomics specialist" in a lab coat. Its Italian edition called her a giornalista, a legally protected title in Italy. And under the EU AI Act, AI-written text can skip the disclosure label if a human reviewed it, while the byline, as Puliafito writes, "belongs to a fictional persona who cannot be sued for defamation." Putting that loophole aside, I think this is a bad idea for anyone looking to build credibility for their publication!

  • Only 12% of people across nine countries are comfortable with news made entirely by AI, according to the Reuters Institute's new Generative AI and News Report — rising to 19% with some human oversight. Meanwhile, in the six countries it has tracked since 2024, the share who used AI to get information in the last week went from 11% to 38%. Lead author Felix Simon walks through it at Nieman Lab. People are happy to use AI to find things out — they just don't want it doing the reporting. (Someone should tell Aria Bloom.)

  • Ogilvy and CAA launched Echo, which lets CAA's actors, athletes and creators license verified AI doubles of themselves for ads, as Digiday's Sam Bradley reports. "Talent always gets to see everything and approve everything at the end," says Ogilvy's Carol Reed, and the ads will carry a watermark marking them as AI-generated and approved by the star. It's still in pilots, with no clients named, and Digiday says it's expected to cut cost-per-asset by 65-75%. At least this one puts a real person's name — and sign-off — behind the AI.

WORK WITH ME
🤝 Want This for Your Team?

I spent 15+ years running digital and AI-adoption strategy inside global brands — the kind of enterprise-scale channel work most "AI consultants" have never actually done from inside a company that size. I bring that same playbook to comms teams directly now: workflow audits, executive AI systems, the intranet-and-channel rebuilds I used to run in-house.

A handful of clients at a time. Reply and tell me what you're stuck on.

SPONSORED BY

What Creators Really Want. Straight From 300+ of Them.

This report shares insights from 300+ creators on what makes brand partnerships last.

  • Learn why 45% of creators want significant or complete creative control

  • Learn why 51% of creators need multiple collaborations before committing

  • Discover what creators say makes them commit long term

Build partnerships that renew instead of reset.

COOL AI TOOLS
🔨 Tools To Try This Week

  • Vellum — Evaluates, tests and compares LLM performance for teams. A fitting pick for a week about checking AI's work.

  • Précis — A PR productivity and content platform built for communications pros.

  • Migma — Creates on-brand, AI-designed email communications in minutes.

  • Canva AI — Canva's design-first AI for visuals, copy and brand assets.

AI + COMMS JOBS
🏢 Find a New Gig

Looking for a role at the intersection of communications and AI? Here are some opportunities to check out:

Was this forwarded to you by a friend or colleague? Want to get this in your inbox next week? Click here to subscribe!

YOUR FEEDBACK WANTED
🔊 Help The Comms Stack Improve

Quick question: how can I help?

What workflows are you struggling with? Where does AI still feel mysterious or overwhelming? What has worked that you’d like to share with others?

I’m a builder, and I’d love to help you and the rest of The Comms Stack community find great new ways to use AI.

Reply and tell me.
I read every response.

Even a one-sentence reply helps. For example:
“I wish AI could help me with ______.”

Until December 2,

Dan