The Feed Report: Issue #4
AI is knocking at your front door and your back door, courtesy of Meta.
Cannes wrapped last week, and every platform showed up to the beach with the same AI party favors. Reddit brought AI ad tools. YouTube brought Gemini. Meta brought a whole rebuilt Creator Studio. If you felt like your feed suddenly filled with “we’re bringing AI to X” announcements, that’s why. The festival is a launch pad now, not just an awards show.
One thing worth naming before we start: Meta is wedging AI into both ends of your workflow this issue. The stuff that reviews your ads before they run, and the stuff that answers your comments after they post. Keep that in the back of your head as you read. It’s less “three separate Meta stories” and more “one platform’s AI strategy showing up in three places.” This AI bubble is getting closer and closer to popping every week.
Burn Notice
The headlines you should know exist. None of these need you to open Ads Manager today.
Reddit opened split testing to every advertiser. Ad split testing is now generally available and self-serve in Reddit Ads Manager, no rep required. You run two variants with one variable changed, Reddit splits it at the user level, and a winner gets called at 65% confidence over a two to six week window. Reddit says four of five early tests found a winning variant on ROAS. If Reddit is in your mix, this is a real upgrade. If it isn’t, file it and move on.
Reddit also dropped a full AI ad suite at Cannes. Beyond split testing: a beta free-form ad generator that builds ads from your site plus relevant Reddit threads, “tailored creative assets” that suggest communities and headlines and image variants, and Shopping Listing Ads in alpha (a multi-advertiser carousel matched to conversations). All running on what Reddit calls Community Intelligence. Alpha and beta, so this is awareness only for now.
YouTube shipped Gemini-powered insights tools. Trending insights inside Google Ads Insights Finder (US), integrated brand pulse metrics, a new Content and Creator Insights API for agency media planning, and Gemini creative tips coming to Demand Gen. Relevant if you run YouTube or Demand Gen, or plan creator media at an agency.
Google finished its June 2026 spam update. It rolled out June 24 to 26, global, all languages, no new policies. This is a standard SpamBrain tune, not a core update. It’s an organic search item that sits at the edge of our lane, but if you or a client watches rankings, it’s on the record.
Google Ads API is sunsetting new Smart Campaign creation. Google’s Ads Developer Blog flagged (June 23) that you’ll no longer be able to spin up new Smart Campaigns through the API. Narrow. This matters to developers and tools built on the Google Ads API and anyone creating Smart Campaigns programmatically. Technical heads-up, nothing to do.
LinkedIn is testing topic-based feeds. Alternative feeds organized around topics, recommended by engagement and trending news, sitting alongside your main feed. LinkedIn trialed it with Cannes attendees first and early feedback is positive. This is attempt number two (last year’s version fizzled), now riding the LLM feed-ranking overhaul from March. If it ships broadly, topic-specific LinkedIn content gets a distribution surface beyond your own network. Not actionable yet, but worth a revisit for organic strategy if it graduates.
The Safe
A soon paid-only exclusive. Where the news gets translated into what to actually do about it.
1. Meta built booking directly into Lead Ads
Here’s the thing that kills more leads than bad creative: the gap between the form-fill and the actual booked appointment. Someone raises their hand, then goes cold in the twelve minutes it takes for anyone to follow up. Meta just went after that dead air.
What happened. Meta launched Embedded Appointment Booking for Facebook Lead Ads Instant Forms. After a lead submits the form, a scheduling widget loads right on the Thank You page. Their name, email, and phone auto-carry over, they pick a slot, they confirm, and they never have to leave Facebook.
Confirmed booking partners at launch are HighLevel and Calendly, the two calendar tools with the widest adoption among lead-gen advertisers. Rollout is phased through HighLevel: 25% of advertisers as of June 24, 50% on August 10, full availability in November 2026.
Why it changes more than convenience. A booked appointment is a stronger optimization and reporting event than a form-fill. A form-fill is a maybe. A booking is a maybe who picked a time. Once you can set the booking as your tracked conversion, you can optimize toward the thing that actually correlates with revenue instead of the thing that correlates with a curious thumb. This minorly redefines in what you report as a “conversion,” and it’s worth thinking through before you flip it on.
Mind the gap. It’s Facebook-only at launch. If you’re running Instagram-heavy lead campaigns (and a lot of you are), those are excluded for now. Don’t rebuild your whole funnel around a feature half your placements can’t touch.
The catch nobody’s talking about. An in-ad booking becomes a second source of truth. If the booking lives in Meta and also syncs to your CRM, you now have two records of the same appointment and a real risk of double-counting or orphaning a lead somewhere between them. Confirm your sync before you scale, or your reporting gets fuzzy fast.
Your next moves:
Check whether your accounts are in the 25% rollout group yet.
Audit which clients can benefit. Anyone selling a slot in a calendar: consultations, demos, tastings, private dining, qualifying-service based opportunities.
Pick your booking partner before you build. HighLevel or Calendly is a stack decision, not an afterthought.
Set the booking as a tracked conversion and decide your optimization target: booking or form-fill.
Confirm CRM sync so you don’t end up with orphaned records.
Still unknown: When it will arrive and how it looks within Instagram, the full partner list beyond HighLevel and Calendly, exactly how the booking event flows into optimization, and how no-shows are handled.
2. Meta is moving 90% of ad and content review to AI
This is the one to pay attention to, because it changes something you touch every single week: ad approvals.
What happened. Per a Financial Times report (around June 25), Meta is accelerating a plan to move the majority of both content and ad review to its own large language models. AI handles roughly 50% of that work today. The target is 90%-plus by the end of 2026 for certain content types, covering policy violations, spam, and ad compliance checks across Facebook and Instagram.
Meta had previously signaled a slower, more human-in-the-loop transition. This is the timeline getting pulled forward, and it lands right after Meta cut around 8,000 jobs (roughly 10% of staff). For extra flavor, a recent exploit compromised about 20,000 Instagram accounts by manipulating Meta’s AI support bot. So that’s the system your appeals may soon route through.
What it means for you. Your disapprovals are about to change character. When AI gets it right, approvals get faster, which is great. When it gets it wrong, you get more baffling false rejections and a thinner human appeals path to fix them. There’s no toggle for this. Your only real lever is the creative itself. Tighten your copy and claims now, especially anything an AI will read literally. Vague-but-fine phrasing that a human reviewer would’ve waved through is exactly what a literal model flags. If you write like you’re being read by a machine that doesn’t do nuance, you’ll trip fewer wires. Health, wellness, before-and-after, and claims-heavy verticals (looking at you, med spa accounts) should get a copy audit first.
The adjacency angle. AI isn’t just reviewing your ad, it’s reviewing what runs next to your ad. If the model gets less precise about what content it allows on the platform, your brand-safety and adjacency risk goes up alongside it. Worth flagging to any client who cares deeply about where their name appears.
Start documenting more. Fewer humans in review means fewer humans in appeals. Your appeal on an AI disapproval may itself be judged by AI. Document everything. Screenshot the ad, save the copy, keep a paper trail on anything you’d want to contest, because “let me talk to a person” is getting harder to do.
Your next moves (this is a monitoring story, not a restructure):
Watch your disapproval rates over the coming months and note any spike.
Pre-empt sensitive verticals now: audit copy and creative for claims a model reads literally.
Keep a documented trail on anything likely to get flagged.
Do not restructure your strategy yet. Watch and prepare.
Get ahead of it with clients. Anyone in a sensitive category will see more disapprovals before they understand why, and they’ll assume you broke something. Say it first: “Meta is shifting ad review to AI. Expect more false flags on health and beauty claims over the next few months. We’re tightening copy proactively so we trip fewer of them.”
Still unknown (and there’s a lot here): which content types hit 90% first, whether ad review is in the first wave or lagging behind content moderation, what the appeals path actually looks like after the transition, and the real false-positive rate. Hold your conclusions loosely.
3. LinkedIn will now write your job applications for you
LinkedIn has a philosophy problem, and it’s kind of funny to watch.
The double standard. Back in May, LinkedIn cracked down on AI “slop” in the feed, positioning itself as the platform that wanted real human posts. Now it’s rolling out Premium Apply Assistant, which drafts your cover letters, pre-fills your applications, and hands you an AI-authored submission. So AI content is a problem in the feed but a premium feature in the application flow. Pick a lane.
What it does. For Premium Career subscribers (a ridiculous $39.99/month), the assistant surfaces high-match roles (including some not posted on LinkedIn), pre-fills application fields from your profile, drafts tailored cover letters, and attaches a confidence indicator scoring how ready each element is. It works on some supported external job sites too. You still review and manually submit. Here’s the key part: LinkedIn says recruiters see only the final submitted content, with no indication AI was involved.
Why this matters even if you’re not job hunting. When applications are AI-authored and undetectable, the early-stage screening signal collapses. A tailored cover letter used to mean someone cared enough to write one. Now it means someone has a subscription. Hiring shifts toward the things AI can’t fake yet: portfolios, in-person conversations, actual work samples. If you hire, weight those heavier and expect application volume up, signal down.
And the flip side. If applications become commodities, your actual LinkedIn presence becomes the thing that separates you. The content you post, the work you show, the way you talk about your field. That’s the part a confidence indicator can’t generate for you. For anyone building a personal brand on the platform, this is one more reason owned content beats a templated application.
Your move: mostly awareness. If you hire, adjust your screening toward portfolios and interviews. If you build a presence on LinkedIn, double down on original content, because it just became the differentiator.
Still unknown: whether this expands beyond Premium, whether recruiters eventually get an AI-disclosure signal, and how often people actually submit the AI draft unchanged versus editing it.
4. Most Americans want under-16s off social media
The headline is the ban. The story for your accounts is age verification.
What the numbers say. A Pew survey (fieldwork late May into early June, roughly 9,750 respondents) found 56% of U.S. adults support banning under-16s from social media, about 20% oppose, and roughly a quarter are unsure. It’s bipartisan, with more people on both sides in favor than opposed, and it spans age groups (young adults are the least supportive, parents the most). Related measures poll even higher: around 78% support time limits for minors, up from 69% in 2023, with strong support for age verification and parental consent. The poll landed two days after the U.S. House passed the Kids Internet and Digital Safety Act, 267 to 117. Australia already banned under-16 use. The UK and Denmark and others are moving on age restrictions too.
Why the ban isn’t the point. A ban requires knowing how old people are. That means age verification, and age verification is what actually hits your accounts. It shrinks your addressable audience to verified adults, adds friction to the funnel, and worsens the signal loss you’re already fighting. Fewer verified-adult impressions can move your CPMs. That’s the real downstream effect, not the headline.
Platforms move before laws do. Teen accounts with default restrictions already exist. Expect more of them, and expect defaults that quietly suppress organic reach to younger cohorts regardless of what any bill does. Platforms would rather build the guardrails pre-emptively than get told how to build them.
The question your clients will start asking. Marketing to teens is reputationally hotter now. Even where it’s perfectly legal, “should we” is becoming a live client conversation, not a settled one. If you have youth-adjacent brands, be ready for that question before it’s asked.
Your move: monitoring. Audit whether any client depends on under-18 reach and scenario-plan if the answer is yes. Don’t restructure anything. Track two trigger events: the Kids Internet and Digital Safety Act’s progress and platform age-verification rollouts. Those are your signals to act.
Still unknown: whether the House bill becomes law and in what form, how U.S. platforms would implement verification if required, and the real audience and CPM impact of verification gating.
5. Meta brought back Creator Studio, now with an AI that answers your comments
And here’s the other end of Meta’s pipeline. Story 2 was AI reviewing what goes in. This is AI handling what comes out.
What happened. Meta is reviving Facebook Creator Studio (which it killed in January 2023 for Business Suite) as a standalone AI companion app. Limited testing, waitlist open. The core is an AI Creator Assistant, an always-on chatbot that answers performance questions (”when should I post?”, “what are people saying in my comments?”), surfaces daily priorities, and tracks goals. The most consequential piece: AI-powered comment management that flags priority comments and drafts replies in your writing style for you to review and edit before posting. Meta is also splitting the Professional Dashboard into separate Creator and Business Dashboards over the coming months and bringing more Business Suite features (content calendar, bulk upload, expanded analytics) to desktop.
What changes about the work. Community management stops being “do the work” and becomes “approve the AI’s work.” That’s a real time saver on high-volume pages. It’s also a voice-and-compliance risk the second you’re managing a branded or sensitive account. An AI drafting replies “in your style” is fine until it confidently answers a customer complaint or a health question in a way you’d never sign off on. Guardrails first, scale second.
The skeptic’s note. The entire point of social is connecting with the people who show up. Outsourcing that connection to a bot is a strange thing to automate. Fast, sure. But you’re automating the one part that was supposed to be human.
The practical heads-up. The Creator versus Business Dashboard split is a “where did my tools go” situation waiting to happen. If you manage brand pages, some of your tooling is about to move. Know that before you’re hunting for a button mid-task.
Your next moves:
Join the waitlist so you can evaluate it firsthand.
Set an internal policy on AI-drafted replies for client pages before you enable anything.
Note the dashboard split is coming so the tool relocation doesn’t catch you off guard.
Flag for clients with managed pages: AI-drafted replies carry brand-voice and compliance risk. An approval workflow isn’t optional here.
Still unknown: wider rollout timing, whether the assistant covers brand and business pages or creator profiles only, and when the dashboard split actually happens.
That’s the week. Three of those Meta stories aren’t three stories, they’re one company deciding AI should touch your work at the front door and the back. The booking feature is the friendly face of it. The 90% review shift is the one that’ll actually affect us.
See you next Thursday.
xoxo,
Mckenna