This website uses cookies

Read our Privacy policy and Terms of use for more information.

In partnership with

Your weekly guide on everything AI marketing. Covering everything from AI marketing news, tips, deep dives, events, podcasts, jobs and much more. Never miss a beat with our 5-minute newsletter.

In today’s email:

  • Deepfaked creators are selling products they have never touched

  • The AI visibility trade is emailing publishers under invented names

  • Sport Clips has 3,400 AI agents answering reviews across its salons

  • Plus: a Bain partner on where AI pressure really comes from, an analytics lead on what not to build with AI, and is retargeting dead?

The Top 3 Stories

1. Deepfaked creators are selling things they have never touched

Scammers are stitching creators’ old photos into fake sponsored posts. Emily Schuman, who has more than 500,000 Instagram followers, turned up in a fabricated ad for telehealth firm Gala holding a vial of GLP-1 drugs. Surfshark research puts 2026 deepfake scam losses as high as $3.7bn, roughly half through social impersonation.

Why it matters: Your influencer contracts assume the creator controls their own face, so the next one should say who watches for fakes and who removes them. 📰 The Guardian

2. The AI visibility trade is emailing publishers under invented names

A firm called Noble automates outreach asking publishers to edit a client’s name into articles they have already published, and sends those emails under made-up names. Its response rate runs between 4% and 10%. One ad tech executive pulled out after two weeks, saying he did not want to tarnish his brand with spammy outreach.

Why it matters: Hire someone for AI visibility and you inherit their methods, because it is your brand name in that email, not theirs. Vet the process, not just the reporting. 📰 AdExchanger

3. Sport Clips now runs 3,400 AI agents across its salons

Sport Clips has deployed more than 3,400 SOCi Genius Agents across nearly 1,800 locations, where they handle over 7,000 customer reviews a month. Each is trained on brand guidelines and tone, then adapts to what is happening at that particular salon. New sites get covered without new headcount.

Why it matters: Replying to local reviews was always capped by headcount, which is no longer much of an excuse for leaving them unanswered. 📰 Chain Store Age

More trending AI marketing news from last week

  • Anthropic’s doom warnings may be “PR and marketing”, a UN adviser said.

  • Estée Lauder handed AI tools to all 14,000 of its employees.

  • Chewy expects roughly $50m in annual AI savings by 2027.

  • Shipt will now build a family dinner basket to a stated budget.

  • Boll & Branch set playbook rules for AI-generated campaign imagery.

Is Your Training Data Actually Model-Ready?

If you're fine-tuning a speech model, you've probably hit this wall: DNSMOS gives you a score, but it doesn't tell you whether the data behind that score is actually right for your model. 

Treat it as a pass/fail gate and you'll end up training on audio that looks clean on paper but drags down real-world performance—while good source data gets tossed for no reason.

Voices' CTO DJ Jalali (with the team's senior audio and voice data engineers) just published a free white paper that breaks down the four-step calibration framework they use internally to set model-specific quality thresholds instead of trusting the raw DNSMOS number. It also covers where DNSMOS breaks down and how Voices validates audio for custom datasets at scale.

Quote of the week

❝

Everybody and their sister has to have an answer to the question of ‘what are you doing with AI’ for their board, for their investors.

Aaron Cheris, partner and global head of the retail practice, Bain & Company, speaking to Retail Dive, 9 September 2026.

Trending AI tools for marketers

Here are 4 trending tools to explore this week. Quick, practical upgrades to your AI marketing arsenal:

  • Wisry: Rebuilds proven ad formats in your branding without briefing a designer.

  • Anysite.io: Builds enriched B2B prospect lists from a plain-language request.

  • Relaticle: Updates your CRM records automatically while you approve every change.

  • Live Captions by Subanana: Lets event audiences follow your talk in their own language.

TV, podcasts & streaming

What AI actually replaces in data work

Julie Beynon of Augment Code cloned her best analytics engineer into an agent that ships a whole pull request from one Slack message. Her test before building anything is who maintains it in six months. You can burn $1,000 rebuilding a tool that costs $5, listen to it here.

AI training

How to build a week of ad creative using Wisry

The overview:

Wisry reads your store, finds ad formats already working in your category, and rebuilds them as static and video ads in your own branding.

Step-by-step:

  1. Point it at your store
    Paste your store URL so the agents pick up your products, palette and tone.
    Skip this and every draft comes back looking like a template.

  2. Let it read the ad libraries
    The research agents scan public ad libraries for formats already running in your category.
    You are looking for patterns, not one clever execution.

  3. Check what is saturated
    Open the research view and read variant counts and days live.
    A format ten rivals have run all quarter is one your audience has learned to ignore.

  4. Rebuild one format as your own
    Pick a single format and have it rebuilt with your product, script and footage.
    One format done properly beats six half-finished variants.

  5. Fix the draft in chat
    Tell the agent what is wrong in plain words and let it revise.
    The makers describe the first pass as a strong draft, so treat it as one.

Pro tip: Keep the formats that perform in a running shortlist, so next quarter starts from evidence rather than a blank brief.

The weekly deep dive

Your retargeting pool is filling up with robots

Bot and AI-agent traffic has stopped being a publisher problem and started showing up in advertisers’ performance data.

The pools got poisoned
GoFish president David Dweck says the agency’s e-commerce clients have seen bot traffic rise 80% year on year, and their retargeting tools were quietly funnelling spend towards visitors who were not people. Costs per acquisition spiked. Performance started degrading in Q4 last year and dragged into this one.

Why the old filters missed it
Bot filters were built to catch things that land on a page and bounce straight off. AI agents add items to carts and sign up to newsletters, so they read as engaged. Mallory Beck of Lighthouse Creative notes this inflates impressions and conversions, muddies lookalike audiences and eats into frequency caps.

The fix is classification, not walls
Narrowing retargeting parameters worked, at the cost of a 20% higher average CPM. Some advertisers gave up on retargeting entirely and moved budget into retail media and social. Chad Keller of Mellow Sleep argues blocking everything non-human would cost more than it saves, because AI assistants already send traffic converting several times better than his site average.

Our takeaway: Stop asking whether a visitor is human and start asking whether the action they took is worth paying for twice.

📖 Read the full article at Digiday

One more thing

Get more than the newsletter…

Did you know your subscription includes a free AI Marketers account? It includes:

  • A personalised AI Marketing Audit that scores your team and gives you a 90-day plan

  • Job alerts matched to your role across hundreds of AI-native companies

  • A directory of 270+ AI marketing tools with editorial reviews

  • Saved jobs, events, and tools in your own portal

Claim your free account → aimarketers.co.uk/claim

Reply

Avatar

or to participate