Shoppers hate AI marketing. Marketers don't care.
Welcome to the AI + Marketing Weekly! Plus: Reddit threads outrank your homepage, Elon vs. Sam in court, and Taylor Swift trademarked her voice.
~The AI + Marketing Weekly~
what you need to know this week in ai + marketing (15-second version):
Only 7% of consumers say visible AI in marketing makes them trust a brand more, while 31% say it makes them trust the brand less - meanwhile, 77% of senior marketers plan to shift budget toward GenAI creator content. (eMarketer)
Taylor Swift trademarked her voice and image - filed three trademarks including sound marks for “Hey, it’s Taylor Swift” and “Hey, it’s Taylor,” plus a visual trademark of her stage image. (Variety)
Over 1 in 5 videos YouTube recommends are AI slop - analysts estimate AI is currently 60% hindering brand safety, 40% helping. Coca-Cola’s AI Christmas ad tested well in focus groups and then tanked publicly two years running. (eMarketer)
LLMs increasingly cite Reddit, YouTube, LinkedIn, G2, and Wikipedia ahead of brand-owned content - Microsoft and Apple sites are cited less often than Reddit threads and Wikipedia entries about their products. (MarTech)
Elon Musk testified for three days against OpenAI, claiming they “stole a charity” - OpenAI’s lawyer countered that Musk only sued because he didn’t get full control. Altman watched from across the courtroom. Two of the most powerful men in tech, one Oakland courtroom, hundreds of pages of receipts. (CNN)
Meta and Google’s Q1 earnings showed AI driving real ad revenue - Meta sales up 33% with 8 million advertisers using its GenAI creative tools. Two years in, AI is finally showing up in the line items that actually matter (The New York Times)
B2B organic traffic dropped 33.6% year over year while direct MQLs grew 6% - buyers are researching off-site through AI tools, then returning directly to brands they already trust. 61% of marketers say this is the biggest disruption in 20 years. (eMarketer)
🗞️ top ai + marketing news
the trust crisis: shoppers, slop, and brand safety
Shoppers aren’t impressed by AI-generated marketing
eMarketer | ~3-min read
Only 7% of consumers say visible AI in marketing makes them trust a brand more, while 31% say it makes them trust the brand less. Meanwhile, 77% of senior marketers plan to shift budget toward GenAI creator content.
My take: The gap between what marketers want to do and what shoppers want them to do has never been wider.
AI’s double-edged impact on brand safety
eMarketer | ~4-min read
Over 1 in 5 videos YouTube recommends are AI slop. Analysts estimate AI is currently 60% hindering brand safety, 40% helping. Coca-Cola’s AI Christmas ad tested well and then tanked publicly two years running.
My take: Testing positive in focus groups does not mean it lands in the wild.
The third-party sites shaping your brand in AI search
MarTech | ~5-min read
LLMs increasingly cite Reddit, YouTube, LinkedIn, G2, and Wikipedia ahead of brand-owned content. Microsoft and Apple sites are cited less often than Reddit threads and Wikipedia entries about their products.
My take: Your homepage isn’t the front door anymore but the Reddit thread from three years ago is.
the ai labor question: jobs, layoffs, and CEO warnings
The hidden logic behind AI CEOs’ job loss warnings
Fast Company | ~6-min read
Sam Altman, Dario Amodei, and Demis Hassabis keep predicting massive white-collar job loss. Ben Goertzel argues the warnings double as the investment thesis of the century: if AI takes the jobs, you better own a piece of AI.
My take: Worth remembering when CEOs are the ones telling you their product will end work as we know it.
AI jobs unemployment Silicon Valley
The New York Times | ~8-min read
Ezra Klein argues a mass AI job apocalypse probably won’t happen, leaning on economists who say demand always shifts toward what can’t be automated. The harder problem is partial displacement: a slow restructuring of who gets in the door at all.
My take: The thing that still disappears is the entry-level rung you can’t see in any chart.
7 ways AI is being used at work
Fast Company | ~5-min read
Real examples from teachers grading 100 papers in 30 minutes, a marketing head at HireQuest using Claude to build a franchise dashboard, and a PM using Claude to translate engineering meetings into plain English.
My take: Love seeing the real stuff vs.hype - people quietly doing their jobs better with the right tool open in another tab.
the trial and the ad boom
Elon Musk takes the stand against OpenAI
CNN | ~5-min read
Musk testified for three days in his lawsuit against OpenAI, Altman, Brockman, and Microsoft, claiming they “stole a charity.” OpenAI’s lawyer countered that Musk only sued because he didn’t get full control. Altman watched from across the courtroom.
My take: Two of the most powerful men in tech, one Oakland courtroom, hundreds of pages of receipts. Whoever wins, the founding myth of OpenAI does not survive this trial.
A.I. helps online ad businesses boom
The New York Times | ~6-min read
Q1 earnings showed AI driving real ad revenue at Google and Meta, with Meta sales up 33% and 8 million advertisers using its GenAI creative tools. Investors gave Google more credit because its AI bets translate directly into cloud revenue too.
My take: Two years in, AI is finally showing up in the line items that actually matter. The ad platforms that cracked it first are about to widen the gap.
Anthropic, Blackstone, and Goldman launch a $1.5B AI services firm
The New York Times | ~5-min read
Anthropic is partnering with Blackstone, Hellman & Friedman, and Goldman Sachs on a new $1.5 billion enterprise services company that will embed engineers inside mid-sized companies to redesign workflows around Claude. OpenAI is reportedly building a near-identical structure with TPG and Bain.
My take: This is consulting rebuilt from the model up. Companies that spend six dollars on services for every dollar on software just got a new option.
the marketer’s reckoning
AI moved forward, marketing did not
MarTech | ~5-min read
Eighteen months in, most marketers still use AI like a smarter autocomplete. Open chat window, type request, edit output. Today’s models can research three competitive angles, draft a follow-up piece, identify the best distribution channels, and write tailored intros in one session.
My take: If your AI workflow looks the same as it did last spring, you are leaving years of capability on the table.
Marketing teams aren’t fully embracing AI. Here’s what leaders can do
CMSWire | ~4-min read
86% of C-suite execs say AI usage is required, but only 49% of middle managers reinforce that with their teams. Marketers hear AI in strategy decks but not in their KPIs.
My take: Helpful training can help with this.
How marketing leaders are succeeding in the AI era
MarTech | ~4-min read
Gartner found that 82% of business leaders say their brand and culture must evolve for AI, but only 15% of CEOs see their marketing leader as strongly AI-savvy.
My take: The marketers winning are using AI for insight and strategy, not just content production.
As AI becomes standard for scaling, marketing leaders are adapting
Digiday | ~3-min read
A new Contentful and Benchmarker survey shows AI hasn’t replaced content outsourcing yet, but 70-85% of companies plan to increase AI investment, mostly in workflow automation and analytics.
My take: The era of “AI for blog drafts” is closing. The next round of investment is about decisions, too.
affiliate, inbound, and the channel shifts
Why affiliate marketing still needs humans in the AI era
MarTech | ~4-min read
Affiliate returns 12 to 15 times spend on average, yet many companies still treat it as set-and-forget. The author argues affiliate is not a channel but a portfolio of partnerships that needs constant calibration.
My take: Software alone has never run a good affiliate program and AI doesn’t change that.
FAQ on inbound marketing
eMarketer | ~6-min read
B2B organic traffic dropped 33.6% year over year while direct MQLs grew 6%. Buyers are researching off-site through AI tools, then returning directly to brands they already trust. 61% of marketers say this is the biggest disruption in 20 years.
My take: Inbound isn’t dead. The first half of the funnel just moved into ChatGPT. If your brand isn’t getting cited there, you’re not in the consideration set.
the celebrity and AI’s edges
Taylor Swift trademarks her voice and image
Variety | ~4-min read
Swift filed three trademarks: two sound marks for “Hey, it’s Taylor Swift” and “Hey, it’s Taylor,” plus a visual trademark of her stage image with a pink guitar. The move follows Matthew McConaughey’s eight similar trademarks last year.
My take: Right of publicity is state-by-state. Federal trademark law is national. If this works, every famous voice and face is about to get the same treatment, and the AI labs are going to have a much harder time pretending it’s all fair use.
👩💻 thought-leader highlights
Poonam Soni – “Teach it once. Claude follows it every single time after”
Poonam highlights Anthropic’s 33-page guide on building Skills for Claude - folders containing SKILL.md files that teach Claude specific workflows for standardizing writing styles, research methodologies, or code reviews. The core benefit: “Teach it once. Claude follows it every single time after.” Commenters noted this signals a shift from “prompt engineer” to “system architect” era, emphasizing the need for execution boundaries and governance checks.
Emily Kramer – “New marketing roles for the AI age: Gen Marketer and beyond”
Emily identifies emerging roles breaking traditional GTM silos: Chief of Staff Marketing, Agentic Operator, and Gen Marketer. The most valuable skill isn’t traditional marketing or engineering - it’s wiring up MCPs, evaluating outputs, and explaining business cases. New titles fail without structural change, and modern CMOs now spend half their time on agent specs and prompt repos.
OpenAI – “Prompt guidance for GPT-5.5, GPT-5.4, and Codex”
OpenAI provides model-specific prompting strategies. GPT-5.5 works best with shorter, outcome-first prompts and explicit retrieval budgets. GPT-5.4 emphasizes multi-step reasoning with clear defaults for when to proceed versus ask permission. Key themes include distinguishing working steps from user communication, grounding answers in evidence, and adding verification loops before high-impact actions.
Axelle Malek – “Essential Claude shortcuts and commands for productivity”
Axelle shares practical slash commands that transform Claude into a structured workflow platform. Shortcuts include /cost for token estimation, /STEP-BY-STEP for action plans, /resume to reopen conversations, /branch for alternate paths, and /memory for persistent notes. Core insight: “Shortcuts exist so you spend less time managing the tool and more time using it.”
Ben Miller – “Tech debt is coming for marketers”
Ben created a /marketing-engineer Claude skill to audit projects and plan systems with better architecture. He discovered problems stemming from architectural flaws - misaligned file structures, overloaded LLM nodes, and uncontrolled vocabularies. Ben warns “tech debt is coming for marketers” as AI building prioritizes speed over structure, predicting the job market will reward genuine comprehension over artificial outputs.
Michael Crist – “How to build an AI Chief of Staff”
Michael introduces Claude Cowork for delegating tasks rather than chatting. The 20-minute setup involves creating a folder structure and writing context files. His “Director’s Brief Framework” structures tasks with Role, Goal, Task, Context, and Constraints. He demonstrates organizing 28 receipts into folders with renamed files and CSV summary in 6.5 minutes—work that typically requires manual effort.
🛠️ latest ai + marketing tools
Claude for Creative Work
Anthropic launched connectors for nine professional creative tools including Adobe, Blender, Autodesk, Ableton, and Splice. Claude now works inside software creatives already use, automating batch tasks, writing scripts, and executing actions directly inside Photoshop, Premiere, and Blender. Also joined the Blender Development Fund and is partnering with RISD, Ringling, and Goldsmiths on creative computation programs.
Meta Ads AI Connectors
Meta opened its ad ecosystem to third-party AI tools, letting advertisers manage campaigns through Claude or ChatGPT via an official MCP server. No API setup, no code. Connect once via OAuth and ask your AI of choice to pull performance data, build campaigns, manage catalogs, or diagnose tracking in plain language.
Hightouch raises $150M at $2.75B valuation
Less than 15 months after its last round, Hightouch closed a $150M Series D led by Goldman Sachs and Bain Capital Ventures. The platform lets AI agents run on a brand’s own customer data to plan and execute campaigns across email, ads, SMS, and web. 100%+ growth for two years running. Clients include Domino’s, PetSmart, DraftKings, and Whoop.
Mistral Medium 3.5 + Vibe Remote Agents
Mistral launched a new 128B open-weights model and moved its Vibe coding agents to the cloud, so developers can kick off tasks and step away. Agents run asynchronously, connect to GitHub, Jira, and Slack, and deliver finished pull requests rather than requiring step-by-step oversight.
Google AI Max goes bigger
Google expanded AI Max with three updates ahead of Google Marketing Live: AI Brief (guide campaigns in plain language), AI Max for Shopping (pull from Merchant Center feeds), and unified travel campaigns. Dynamic Search Ads are being retired in favor of AI Max by September.
April Gemini Drops
Google’s April update added a native Mac app, personalized image generation using your Google apps context, free music creation via Lyria 3 Pro, and NotebookLM built into Gemini. Personal Intelligence, connecting Gemini to Gmail, Calendar, and Drive, is now rolling out globally.
Creo (Omnicom) + Google Cloud
Omnicom’s influencer agency Creo launched an AI tool that automatically edits creator content for brand compliance before it goes live. Built on Gemini and Veo, it removes restricted objects, blurs competitor logos, and adjusts wardrobe elements in minutes rather than days. Live in the US now with global rollout in Q3.
Opal Gem
Marketing planning platform Opal launched Gem, an AI copilot built to reduce the “alignment tax”: time spent on status updates, data-chasing, and cross-team coordination. Trained on a brand’s existing Opal data, runs on Claude by default, and replicates past campaigns and generates briefs without manual uploads. Clients include SAP, Boeing, and Starbucks.
Replit Slides
Replit launched an AI slide builder that creates full pitch decks from a text description, with pixel-perfect exports to PowerPoint, Google Slides, and PDF. Start from scratch, import an existing PPTX, or upload brand guidelines, then iterate in chat. Available for Core and Pro users now.
Echo-2 (SpAItial)
SpAItial released Echo-2, a model that turns a single photo or text prompt into a navigable 3D world viewable in any browser in real time. Unlike video generators, it produces a spatially consistent scene you can move through. Built for digital twins, real estate, game prototyping, and robotics simulation.
💼 cool ai + marketing jobs
Serval - Agentic Operator, Growth Marketing
Perplexity - Demand Generation Lead
Scale AI - Product Marketing Lead, GenAI
Profound - Marketing Engineer
OpenAi - Product Marketing Lead, Advertising
Product.ai - Founding Lead of Growth — Brand & Distribution
Reply anytime and let me know:
Topics you’d like covered in future weeks
Sections you love or could do without
Questions about AI, marketing, or anything else
Love y’all,
Carley