The CommunicAItor's Digest
Helping you stay up-to-date with the impact of AI on communications
As of yesterday, May 28, over 1,000 people now subscribe to this digest on LinkedIn and another 400+ on Substack. While this isn't about the numbers per se, the headlines this week relating to white-collar redundancies due to AI are a stark reminder of just how important it is that we stay on top of this, build crucial new skills, and push the boundaries of what comms can achieve with a creative mindset and some sharp AI knowledge!
As Amy Stettler said so brilliantly this week, it's time to “design for the system, not just operate within it.”
For those of us in the job market, I hope the coming week brings you some joy. Updated #Commsjobs links are at the bottom. Enjoy!
Content Creation
Flow + Veo 3 Shrink a Film Crew to One Prompt
What happened: Google’s new Flow timeline editor now ships alongside Veo 3 inside Gemini. Feed the pair a paragraph and within ~90 seconds you get a 30-second, 1080p video complete with color-grade LUTs, ambient score, auto-captions and optional voice-over. Pilot brands (a global sneaker retailer, a cruise line and two tier-one agencies) say they’ve collapsed multi-day shoots and four-figure budgets into lunch-break prototyping. Flow exports directly to YouTube, Drive and Premiere, signaling Google’s bid to own the entire stack from ideation to distribution.
So what for communicators: Storyboarding, rough-cutting and approval can now live in the same afternoon sprint. Re-factor timelines, route that freed cash into media spend or variant testing, and brief stakeholders that “AI turn-time” means feedback windows must tighten. Source
Stitch Auto-Designs Your App UI — and Hands You the Code
What happened: Stitch, a Google Labs experiment running on Gemini 2.5 Pro, converts plain-language prompts or a phone snap of a whiteboard into polished multi-screen mocks plus exportable Figma files or responsive CSS/HTML. Live on Google Labs, it supports theme tokens, variant testing and WCAG checks. At I/O a Googler rebuilt a fintech dashboard in three minutes; a beta design shop reports 70 % of new briefs now start with a Stitch draft.
So what: Slide-deck comps are about to feel prehistoric. When execs realize UI can be prototyped live in the room, they’ll want clickable demos, not flat PDFs. Bone up on “design-sprint-in-an-hour” facilitation and lock brand fonts, color tokens and legal boilerplate into Stitch’s reusable presets. Source
Animon.ai Brings One-Click Anime to Korea
What happened: CreateAI’s Animon.ai launched in Seoul with Hangul and Traditional Chinese UIs plus an unlimited-outputs subscription that undercuts token models. Local studios are already pumping out daily TikTok-length anime serials; K-pop labels test AI-generated “mini-episodes” between music drops. Tech upgrades include frame-consistent character rigs, full-scene relighting and Shorts/Reels presets.
So what: Vertical-specific generators trained on vertical-specific content are splintering the once-monolithic gen-AI stack. Expect each business unit (sports, fashion, travel) to adopt its own specialist tool. Build a cross-platform style bible now so logos, licenses and disclaimers stay consistent. Source
Influencers
Gen Z to Brands: ‘Bots Count as BFFs’ (46 % Prefer AI Creators)
What happened: A Fast Company/Whop survey of 2,001 U.S. Zoomers shows nearly half prefer brands that collaborate with virtual influencers; 33 % say they usually convert after an AI avatar’s rec. Authenticity now skews toward meme-speed humor and transparent labelling over polished studio shoots.
So what: Sounds crazy but build a hybrid roster: synthetic avatars for always-on reach and human micro-creators for BTS and live Q&A. Track trust as closely as clicks—disclosed AI collabs can actually lift sentiment among under-30s. Source
OnlyFans Founder Bets $150 M on “Subs,” an AI-Savvy Creator Hub
What happened: Tim Stokely’s Subs platform goes after mainstream influencers with AI-generated highlight reels, price-optimization tips and long-form “Shows.” Wired pegs the creator-economy TAM at $500bn by 2027. Subs’ AI toolkit aims to grab a slice by automating tiering, churn prediction and content recs.
So what: Platform churn is getting cheaper. As AI lowers switching costs, structure contracts that travel with the talent, and insist on data-share clauses so you can benchmark performance across whichever platform they jump to next. Source
Typeform Warns of a Credibility Crunch (35 % Distrust AI Posts)
What happened: Typeform’s 1,300-respondent study finds a vocal minority doubting unlabeled AI content; posts without clear disclosure saw 2× engagement drop-offs and a 3× spike in negative comments.
So what: Transparency is now an ROI lever. Slap a subtle “Created with Virtual Creator Studio” tag on synthetic posts and pair them with authentic lo-fi footage to keep trust metrics healthy. Source
Executive Communications
Fleishman Hillard builds a 50-Counsel AI Advisory for the C-Suite
What happened: The agency’s new Global Executive Advisory links senior counsellors, data scientists and an AI-risk scorecard to guide leaders through IPOs, activist raids and geopolitical flashpoints. Services range from avatar-message testing to 20-market sentiment heat-maps.
So what: Boards or finance may soon ask, “Where’s the AI risk map?” before green-lighting talking points. Budget for AI stress-tests alongside media coaching, and prepare data-backed narrative options, not just message trees. Source
Zoom & Klarna CEOs Debut Digital Twins on Earnings Calls
What happened – Following Klarna, Zoom’s Eric Yuan pre-recorded his opener via Zoom Clips, then took live Q&A himself. Analysts grilled him on disclosure safeguards; Yuan pitched it as a product demo.
So what – Synthetic stand-ins slash rehearsal time but raise Reg FD eyebrows. Draft a policy on when avatars are allowed, how to label them in transcripts and who eats liability for mis-statements. Source
Regulators Call AI Avatars a ‘Patchwork Risk’
What happened: PYMNTS details how the U.S. relies on FTC deception statutes and state impersonation laws while the EU AI Act mandates disclosure. Legal experts warn of liability if avatars hallucinate forward-looking statements.
So what: Before your exec goes digital, align Legal, IR and Comms on a global checklist. What’s fine in New York may need extra disclaimers in Brussels or Sydney. Source
Internal Communications
Teams Opens Copilot Interaction APIs—Audit Every Prompt
What happened: Microsoft’s new Interactions Export API logs each Copilot Q&A across Teams, Outlook and Word (prompts, answer and citation links) for compliance or analytics.
So what: For the first time you can prove which policies staff actually search for and how they implement them. Bake AI-usage dashboards into your KPI reviews and patch knowledge-base gaps fast. Source
Slack’s Agentforce Bots Join Channels (Plus Shared Folders)
What happened: May release notes let admins drop autonomous Agentforce bots into channels and add 100-item shared folders. Early adopters route IT tickets and HR FAQs to bots, freeing humans for nuanced threads.
So what: Write a tone guide (pronouns, emoji policy, escalation triggers, etc.) before the bots start talking on your behalf. One rogue emoji can undo months of culture work. Source
Ragan Benchmark: AI Adoption Up, Email Still King
What happened: Ragan’s 2025 benchmark shows 72 % of comms teams use at least one AI tool, yet the humble newsletter remains the top channel for reach and recall.
So what: Blend, don’t replace. Use AI for snackable micro-briefs in Slack/Teams, then drive depth via weekly email digests to hit every attention span. Source
Analyst Relations / Investor Relations
Q4 Previews IRO Agent™ Ahead of NIRI 2025
What happened: Q4 will demo a conversational “IRO Agent” at NIRI Boston, promising board-ready answers and real-time activist-sentiment alerts at booth #413.
So what: If your IR stack can’t auto-surface likely activist questions or sentiment swings, expect peers and your CFO to ask why. Time for a tool audit. Source
Box Spikes After AI-First Earnings Script
What happened: Box’s call led with its AI Agents roadmap; analysts loved the concrete KPIs (attach rate, margin impact) and rewarded the stock. CEO Aaron Levie called AI upsells a “growth flywheel.”
So what: Wall St. is past buzzwords. Bake specific AI metrics—ARR uplift, churn delta into your next deck or risk being filed under “hype-no-substance”. Source
UBS Deploys AI Analyst Avatars for Client Videos
What happened: The FT reports UBS is rolling out Synthesia-powered analyst avatars to churn out 5,000 research clips a year. Scripts require human sign-off; avatars are clearly labelled.
So what: Short-form AI video is creeping into high-stakes finance comms. Start auditing voice-rights, IP clauses and disclaimer language before your analysts go synthetic. Source
Crisis Communications
Emotion-Tracking Dashboards Promise 24-Hour Warning
What happened: PR Daily spotlights sentiment tools that immediately identify the intensity and velocity of sentiment spikes (rather than outright volume) which can then auto-compile an issue brief (key posts, media list, draft quotes) on the topic before a spark becomes a wildfire.
So what: Speed beats perfect prose. Marry alerting to pre-approved templates so your first holding line ships in < 60 minutes, not six hours. Source
Deepfake Targets NRLW (National Women's Rugby League) Star; Integrity Unit Steps In
What happened: Gold Coast Titans winger Jaime Chapman revealed AI-manipulated bikini shots; New South Wales Police and the NRL Integrity Unit launched an investigation. Similar attacks hit Australian journalists the same week.
So what: Deepfake abuse is mainstream. Draft a takedown SOP, line up mental-health support for staff and keep pre-written holding lines on the shelf. Source
Muck Rack Publishes an AI Crisis Cheatsheet
What happened: The May 29 guide walks PR pros through using gen-AI to draft initial statements, FAQs and social copy, then pairs each with a human-review checklist.
So what: Treat it as a tabletop-training kit: rehearse AI-speed drafting plus the critical human veto so velocity never outruns accuracy. Source
Disinformation / Misinformation
Meta Axes Fact-Checkers, Bets on Community Notes
What happened: Meta’s Q1 2025 integrity report shows it ditched third-party fact-checkers, narrowed hate-speech definitions and claims a 50 % drop in moderation errors, while bullying and violence ticked up.
So what: With expert gatekeepers gone, brands must self-police harder. Spin up real-time impersonation monitoring and prep rapid rebuttal posts. Source
Google AI Overviews Still Thinks It’s 2024
What happened: Wired caught Google’s search-answer feature repeatedly stating the wrong year and citing fringe sites. Google says such examples “help refine” the model.
So what: Even first-party AI can hallucinate basics. Never quote AI-generated facts without human QA—and consider disclaimers for any AI-assisted copy. Source
Study Finds Deepfakes in Every 2023-24 National Election
What happened: A Steptoe risk briefing, citing Recorded Future, says AI-generated fakes surfaced in all 30 national elections analyzed between July 2023 and July 2024, targeting debates and local ads alike.
So what: Election-year brand campaigns need a manipulated-media lens—even in niche markets. Arm local spokespeople with fact-checking resources and scenario-plan rebuttals now. Source
Available for work?
Andrea Donatucci has a great update here.
Amanda Ponzar has a great list of roles across a number of posts.
Kerri Blanford has a shortlist here.
Leah Rosenfeld has a comprehensive (80+ editions) of comms open roles here.
Marc Goldberg has an updated list here.
Marni Gordon, PCC regularly posts some great roles.
Projects
I built a "Spot the ball" web app in Anthropic Claude Opus 4. In under 20 minutes. I do words - not code, so I have no shame in admitting that this is an incredible achievement!
Aside from the sheer simplicity of the input (I used a very simple prompt), the logic and creative thinking Opus 4 applied (especially in respect of the ball removal and target system) was outstanding. I need a few more tweaks, but an idea that began with a "what if" scenario on Monday turned into an 80% functioning real-world thing two days later.
