the gov unplugged Anthropic's best model 🔌
Welcome to this week's AI + Marketing Weekly! Plus: Publicis coined "AI pitch-maxxing" and made a silly film about it and Cannes kicks off next week with an existential question about creativity.
~The AI + Marketing Weekly~
what you need to know this week in ai + marketing (15-second version):
OpenAI launched product feed campaigns in ChatGPT Ads Manager on June 11, letting retailers upload full catalogs (up to 1 million SKUs) and auto-generate shopping ads. CPC bidding and conversion tracking are now live. (Search Engine Land)
Google Search information agents went live on June 12 for AI Ultra subscribers in all AI Mode languages. The agents monitor news, blogs, social, and shopping data in the background 24/7 and push updates to users when something changes. (PPC Land)
Anthropic announced Claude Corps, a $150 million fellowship placing 1,000 early-career workers inside U.S. nonprofits with $85K salaries and AI training. First cohort of 100 starts October. Applications close July 17. (Anthropic)
The Trump administration ordered Anthropic to suspend access to Claude Fable 5 and Mythos 5 on June 12, citing national security concerns over a reported jailbreak. Anthropic had to disable the models for its entire customer base to comply with the export control directive barring foreign nationals. (TechCrunch)
Publicis CEO Arthur Sadoun released a mockumentary film calling out “AI pitch-maxxing,” his term for agencies overselling AI capabilities to win business, ahead of Cannes Lions. His warning: over-promising on AI is leading to massive job cuts in the industry. (The Drum)
Social commerce crossed $1 trillion globally. TikTok Shop projects $23.4 billion in U.S. sales this year (up 48% YoY), and 76% of people who watched a TikTok Shop livestream bought something. (Ringly.io)
Cannes Lions 2026 opens June 22 with new AI Craft award subcategories, stricter integrity rules after last year’s fake-work scandals, and a growing consensus that the industry needs to prove AI creates business value, not just headlines. (AdWeek)
Thanks for reading and your patience as we publish on Wednesday (again) this week. Happy hump day 🐪 and see you next week!
🗞️ top ai + marketing news
chatgpt becomes a shopping channel
ChatGPT Ads Manager launches product feed campaigns
Search Engine Land | ~4-min read
OpenAI opened product feed campaigns in ChatGPT Ads Manager on June 11. Retailers connect a catalog (up to 1M SKUs), set filters, and ChatGPT auto-generates sponsored placements. It uses the same feed format as Google Shopping, so existing feeds work without rebuilding. CPC bidding and conversion-optimized campaigns are now live.
Why it matters: Running catalog-scale ads on ChatGPT was impractical until now. Test this while CPCs are low and most retailers have not shown up yet!
google search gets a background agent
Google launches always-on information agents in Search
PPC Land | ~4-min read
On June 12, Google launched information agents in Search for AI Ultra subscribers across all AI Mode languages. Users ask AI Mode to monitor a topic and the agent runs 24/7, sending updates with source links as new information surfaces. It crawls news, blogs, social, finance, and shopping data. Expanding beyond Ultra this summer.
Why it matters: Google turned Search into a monitoring service, and the same signals that drive AI Overview citations now determine whether a background agent flags your content to someone tracking your category.
the ai skills gap gets a free fix
Adobe and LinkedIn launch free AI training for marketers
Adobe | ~4-min read
Adobe and LinkedIn launched free AI courses for marketers on June 16: four role-based learning paths (digital marketing, content, social, data) in 47 languages, 2-3 hours each. Available on LinkedIn Learning and Adobe Experience League. AI-skill job postings are up 113% year-over-year, but only 4% of marketers have added AI skills to their profiles.
Why it matters: Always cool to check out new learning courses.
consumers are warming up, with conditions
Invoca finds AI trust improving, but speed still wins the sale
Invoca | ~5-min read
Invoca’s B2C Buyer Experience Report (June 9) shows AI sentiment softening: the share who said AI made buying worse dropped 11 points, and 46% now say it made things better. But 79% will still switch to a faster competitor, 83% want human connection for high-stakes purchases, and 38% blame the brand when AI goes wrong.
Why it matters: Consumers do not care if speed comes from AI or a person. They just want faster answers. And if your chatbot fumbles, your brand takes the hit, not your vendor.
cannes braces for the ai reckoning
Cannes Lions 2026 previews a festival built around proving AI actually works
AdWeek | ~5-min read
Cannes Lions opens June 22 with new AI Craft subcategories and stricter integrity rules that ban agencies up to three years for submitting fake work, after last year’s AI-manipulated case study scandals. Publicis CEO Arthur Sadoun dropped a mockumentary roasting “AI pitch-maxxing,” warning that over-promising on AI is leading to massive industry job cuts.
Why it matters: The industry spent two years saying “look what AI can do.” Cannes 2026 is the moment it has to prove AI did something. The question is shifting from “are you using AI?” to “show me the business result.”
anthropic’s nonprofit bet
Anthropic launches $150M fellowship to embed AI workers in nonprofits
Anthropic | ~4-min read
Anthropic announced Claude Corps on June 11: $150 million to place 1,000 early-career workers inside 400 U.S. nonprofits with $85K salaries, Claude training from CodePath, and weekly Anthropic office hours. No degree required. First cohort of 100 starts October. Applications close July 17.
Why it matters: Nonprofits have big missions and tiny teams, exactly where AI has the most leverage.
anthropic’s biggest week became not so great
U.S. government orders Anthropic to pull Fable 5 and Mythos 5
TechCrunch | ~5-min read
Three days after launching Fable 5, Anthropic had to shut it down. The Trump administration issued an emergency export control directive on June 12 barring foreign nationals from accessing Fable 5 and Mythos 5 over a reported jailbreak. Anthropic could not verify every user’s nationality, so it disabled both models for everyone. This follows the February federal ban after Anthropic refused to allow Claude for mass surveillance or autonomous weapons.
Why it matters: Platform risk is not theoretical. If your workflows depend on a single AI provider, you need a fallback that does not take a week to activate.
👩💻 thought-leader highlights
Allie K. Miller - “You can now build and deploy a live app without writing code”
Allie walks through building and shipping an app using Claude Code for about $20/month. The best line came from the 192 comments: “The code was never the moat. Understanding what people actually need... that’s still the whole game.” The barrier to building collapsed. The barrier to building something people want has not moved.
Nickey Skarstad - “We ran a Claude Design workshop at Duolingo and open-sourced the guide”
Nickey’s team ran an internal Claude Design workshop and published the playbook. The insight worth stealing: “Reserving time on calendars is the ONLY WAY people reliably try new tools.” If you have been trying to get your team to actually adopt AI instead of just talking about it, this is the template.
Jonathan Martinez - “I’m building a private community for AI-focused marketers”
Jonathan is launching a private WhatsApp group for marketers actively using AI, just practitioners sharing what works. 1,000+ comments suggest he hit a nerve. If you are experimenting alone and want a room of people doing the same, he is taking requests.
Tayla Burrell - “13 actually useful things you can do with Claude Cowork this weekend”
Tayla lists one-time projects (email-to-CRM migration, content audit, competitive teardown), recurring tasks (trend briefs, dashboards), and repeatable workflows (content repurposing, Notion builds). Her best line: “Most of these get dramatically better once Cowork understands your context extremely well.”
Barbara & Bojana - “A self-learning AI content system built with Claude”
Barbara and Bojana write how one creator automated his weekly LinkedIn content cycle: pulling signals from news, LinkedIn, and YouTube every Monday at 8 AM, scoring ideas against ICP and brand voice, then drafting posts with dual hook variants. The key detail: the system enforces specific viewpoints during scoring, filtering out flat observations before drafting starts.
Nirnaya - “The room where women cheered for each other”
Nirnaya recaps a gathering where women shared how they are using AI to rebuild after burnout, layoffs, and career resets. One built a mental health app. Another designed her backyard with Claude after a layoff, then built a styling tool. The throughline: AI as recovery tool, not productivity hack.
🛠️ latest ai + marketing tools
Contentstack Agentic Experience Platform (AXP)
An autonomous agent layer for enterprise content, data, and real-time personalization.
Launched June 9 with Agent OS connecting content, data, and AI orchestration so agents manage personalization and activation without manual coordination. Early adopters report 95% less manual reporting. For enterprise teams that need agentic AI in the CMS, not bolted on.
ChatGPT Ads Manager Product Feeds
Upload a product catalog and auto-generate shopping ads inside ChatGPT.
Connect your Google Shopping-format feed (up to 1M SKUs), set filters, and ChatGPT generates sponsored cards below AI answers. CPC/CPM bidding and conversion tracking are live. Existing Google product feeds work without rebuilding.
Google Search Information Agents
Background agents inside Google AI Mode that monitor topics and push updates.
Tell AI Mode what to watch and the agent scans news, blogs, social, and shopping data 24/7, sending summaries with source links. AI Ultra subscribers only for now, expanding this summer.
Raptive Intelligence
AI-native platform turning creator content and audience data into brand-usable signals. Launched June 9 after acquiring AlchemyAI. Packages creator expertise and consumer intent so CPGs, retailers, and AI platforms can access it programmatically. For brands that want creator-ecosystem reach without building the pipes.
Doceree Clinical Intent Signals (CIS)
Real-time clinical intent detection for healthcare marketing. First commercially available intent layer for omnichannel HCP campaigns. Pilot showed 38% faster HCP decision progression and 27% higher engagement vs. control. For pharma marketers running multi-channel campaigns.
💼 cool ai + marketing jobs
Lovable - Solutions Marketing Lead
OpenAI - Head of Ads Enterprise Marketing
Perplexity - Solutions Product Marketing Manager
Guidepoint - Global Head of Marketing & AI Innovation
Replit - Senior Brand Marketing Manager
Anthropic - Head of Copy & Content, Creative Studio
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