Advertise in Geekout
Get in front of 30K social media pros
60%+ Open Rate | 2-4% Ad CTR
Prev sponsors: VEED, Sprout Social, TikTok
Hello, geeks!
It’s been a big week of AI updates. Most of it was Meta being…well… Meta.
Its new Muse Image model, launched earlier this week, can now paste other Instagram users into AI-generated photos — you just @ them — and your public snaps are fair game unless you go trawling through your settings to opt-out. 🙄
Then there are Meta’s incoming "super sensing" glasses that record and capture everything all day long, a now confirmed facial recognition feature for its Ray-Ban smart glasses, and a patent for a new wearable that tracks your emotions and watches you take your meds.
In summary: Meta just cannot stop being… creepy af.
Which is a shame, because AI is extremely useful when it’s not being deployed like a device in a Black Mirror episode with a cheerful gradient logo.
Case in point: I've spent the week down a full vibe-coding rabbit hole, building daft little tools to run Geekout.
OpenAI reached out to offer me early pre-launch access to its new ChatGPT for Work tool last week — and in a few days, it helped me build my own Geekout Intelligence Suite:
-
-
A few of the powertools I built:
🚨 Geekout News Alert Radar — scans every source I follow each morning and ranks the day's social stories by momentum, so I catch the big one before anyone's written it up
-
-
✨ Geekout Subject Line Grader — grades every draft subject line against years of my real Beehiiv open-rate data, spots the winning formula, and predicts the top performer before I send
-
-
💰 Geekout Deals Pipeline Radar — reads my sponsor threads in Gmail and reconciles them against Xero, so it always knows which deals are pitched, agreed, invoiced and paid
-📈 Geekout Sponsor Report Generator — turns raw Beehiiv click data into a polished, client-ready PDF performance report the moment a placement ends
-⭐️ Geekout VIP Subscriber Alerts — taps the Beehiiv API and pings me the second someone notable joins — a big brand, an agency, the odd person from Meta itself 👀
I may package these up as ChatGPT and Claude Skills you can download, drop into your own workflows, and point at your own data.
Interested? Hit reply and tell me which one you’d want first.
Right — Coming up in this week’s Geekout:
🤬 Meta's Muse Image sparks a privacy revolt — opt-out by default, because of course
-📉 Google Search's shiny new tool for social teams — actually useful
-🔢 View counts on Facebook posts are now a thing — next stop, Instagram?
-🚫 TikTok, locked until you've walked 3,000 steps — there's a new app for that
Let’s get into it!
— Matt
__________
3 cool things
👀 I got early access to this wild new AI for social teams → The first AI that didn't make me rewrite everything → Get Early Access Invite* (limited spots)
-
💰 Stop guessing creator collab rates → 7 key factors. 2 real-world formulas. 2026 benchmarks. Free → Grab the Cheatsheet*
-
🤫 The newsletter I quietly STEAL smart insights from → Not an ad — I just rate it. 10 min read. Free → Take a look
*sponsored links
SPONSORED ▶︎ FREE CHEATSHEET
Social Media Managers: Here's what's actually fair to pay creators in 2026 💰
-
Figuring out what to pay creators is one of the trickiest parts of running a program — and it's only got harder as rates climbed as much as 50% this year.
Pay too little = quality drops. 📉
Overpay and scale = quickly over budget. 💸
If you've ever guessed at creator rates or fumbled through a usage rights conversation, this is your cheat code.
Insense just dropped a FREE Cheatsheet that breaks it all down:
7 factors to weigh before setting your creator budget
Real benchmark rates for UGC, organic posts, and whitelisting
2 formulas to build your own organic-posting rate
A cheatsheet to avoid over- or under-paying creators
Built for brands running creators at scale, not one-off collabs.
Grab it now. Avoid the budget regret later!
P.S. Insense helped DTC wellness brand Fling scale from 10 to 60 creators and hit 100K+ organic views — no guesswork required.
⤷ BONUS: Book a free strategy call by July 24th and get $200 towards your first campaign. 👌
Meta builds AI detection tool to ID images and video created with its new models [DETECTING]
Meta tests “super sensing” glasses that capture audio + images all-day-long [EVESDROPPING]
Google Search can now show traffic sent to YouTube, TikTok, and your other social accounts [GOOGLING]
Instagram caught running paid ads promoting child sexual abuse material in India [SICKENING]
SPONSORED ▶︎ SOCIAL AI
EARLY ACCESS: Everyone's AI Posts Sound the Same — Yours Don't Have To! 💡
-
Be honest… you can spot an AI post on LinkedIn (or anywhere else) within two sentences now.
Same structure. Same tone. Same fake-deep one-liners.
96% of marketers use AI for social — but 40% still rewrite every output because it doesn't sound like them.
That's not a shortcut. That's a second job.
Meanwhile you're…
Rewriting the same AI draft for the third time 😭
Watching your brand voice flatten into generic LinkedIn AI slop 💩
Here's the fix: Archie, the AI content engine built for social.
It skips the blank-prompt problem entirely.
I got to play with it…. Here's how it works:
🗂️ Upload what you already have: a webinar, strategy doc, podcast, or whitepaper
🎯 Define your audience, goals, and brand voice once
✍️ Archie extracts the angles worth posting and generates platform-native content for LI, FB, IG & more
🔁 Performance data feeds back into every new content cycle, so results sharpen over time
⚡️ AI-assisted, not AI-generated: it sounds like you from the first draft
4,000+ social teams already use Agorapulse's AI features daily.
Agorapulse is sending out a limited number of beta invites before the public launch.
👉 Skip the queue → Get EARLY ACCESS (limited slots)
Meta can’t stop being creepy
Why does Meta insist on making things so… creepy?
Just this week we’ve seen multiple examples of this decades-old phenomenon.
Meta made headlines this week with the launch of Muse Image across Meta AI, Instagram Stories and WhatsApp. It can generate, edit, restore and remix images, with Muse Video coming soon.
What’s more, advertisers will soon be able to use Muse Image for ad variants and creative generation.
So far, so standard for social media in 2026. But then there’s the crucial bit: users can also @ mention Instagram accounts to pull people into AI-generated images. Meta says public Instagram photos may be used, unless users opt out in the settings.
Yes, this is OPT OUT, not OPT IN. 🤦♂️
Hollywood is already freaking out, calling for the feature to be opt in as movie stars fear exploitation by random Instagram users. Of course, celebrities aren’t the only potential victims here. Regular people have plenty to worry about with how this could be used against their will.
And that’s not all. This week we learned that Meta has patented an AI wearable that could track your emotions all day.
And then there are the “super sensing” glasses that capture audio and images throughout your day, which the Financial Times this week reported as being in the works.
Oh and, Meta CTO Andrew ‘Boz’ Bosworth admitted this week that the company is working on facial recognition features for its AI glasses.
To be fair, he did try to make it sound less dystopian than it first appears. Boz says the feature would only identify people you already know, not random strangers. He said it would be local, encrypted, and won’t pull from a central database.
That makes it sound more like a personal memory aid than public surveillance. But consent is still the awkward bit for people being recognised. And how will Meta know whether someone you already “know” isn’t a stranger you happen to see regularly?
Meta has a history of pushing boundaries around what the public see as acceptable. Facebook users freaked out about the launch of the News Feed feature 20 years ago, but it turned into one of the platform’s most useful features.
But there’s ‘seems a bit creepy at first’ and there’s ‘definitely creepy’, and Meta seems to cross that line all too often.
WANT TO OPT OUT of letting Meta’s new AI image generator create pictures of you without consent? Here’s how.
Meta
/ Facebook
/ Instagram
[Image: @oncescuradu on Threads]
Facebook is testing displaying view counts for posts
Meta is trying to stop people turning its AI glasses into creepware. Its glasses will disable the camera if the recording LED is tampered with
Meta confirmed facial recognition is being explored for its AI glasses
Meta confirmed it’s not given up on VR headsets, and more are in the works
Meta has been grilled over whether “less censorship” unleashed more hate speech
Meta’s in-house Iris chip for AI is expected to enter manufacturing in September
Instagram is testing a new Playback menu for Reels that is home to the closed captions and ‘view full-screen’ features
Instagram’s ‘Add Yours’ Story templates can now be cross-posted to Facebook
__________
Threads
/
WhatsApp
[Image: @oncescuradu on Threads]
Threads now allows users to translate messages
Threads is rolling out the ability to attach a podcast link to a video to more users
WhatsApp is now testing a green dot for online contacts on iOS, as well as Android
WhatsApp is testing birthday notifications for contacts
WhatsApp is testing a new paid partnership label for posts created in broadcast channels
WhatsApp is rolling out notifications for pinned messages
WhatsApp is working on the ability to save a contact’s username in addition to their phone number
__________
X / Twitter
[Image: X]
X is giving creators a proper in-app video editor, featuring multi-language captions and green screen recording
X is testing the ability to share a link to any search
X will soon send you a DM when a post you’ve engaged with is corrected, Elon Musk says… although X already sends notifications about this
Elon Musk will pay just $1.5 million to settle the SEC’s Twitter stock case over his late disclosure of his stake in the company in 2022
A US judge has upheld a jury verdict that Musk defrauded Twitter investors in 2022.
xAI (parent company of X and part of SpaceX) has completed its rebrand to SpaceXAI
__________
TikTok
/
YouTube
[Image: Beehiiv AI]
TikTok is tightening its crackdown on AI spam accounts, with improve detection of accounts mass-posting AI-generated content
TikTok has also launched a drive to help users spot and understand AI-generated content
TikTok can now help brands scale their microdramas with its Growth Max ad product
TikTok has been spotted is now offering updates on trending content and more via SMS
TikTok US launched a six-city ‘Discover America’ tour for America’s 250th birthday, taking its marketing pitch on the road
YouTube now lets eligible creators add 15 seconds of licensed music to image carousels
YouTube is expanding supervised accounts for younger children to more countries
__________
Bluesky
/
DuckDuckGo
Bluesky is introducing an advanced search feature
Bluesky is rolling out the ability to share a link to any search
DuckDuckGo says its browser can now block YouTube ads
Character.AI is jumping on the microdrama content trend with its new c. ai Series format that offers short, vertical, episodic videos on mobile
Instagram has added 30+ new AI effects for Stories.
You can apply one-tap edits like Night Flash, Disposable, Puffer and Paper Doll.
A redesigned effects browser shows previews before applying edits to photos, and you can also write your own prompts to create custom AI effects.
[Image: Meta]
__________
Also spotted this week…
Meta is adding clearer AI labels to ads on Facebook and Instagram
Meta is giving brands better ways to measure AI chatbot performance
Meta launched Muse Spark 1.1, its strongest model yet for coding and agentic work
YouTube is bringing its AI search chatbot to desktop
SpaceXAI launched its new Grok 4.5 model
Google is expanding AI ad disclosures across Search, YouTube
and Discover
Google Photos is getting an AI video remix tool
-
__________
-
-
__________
-
Erling Haaland has made a long-standing meme official.
The Norway striker has long been compared to the Dragon Ball Z villain Majin Buu.
The joke resurfaced lately as Haaland became one of the breakout stars of the 2026 World Cup.
An Instagram post showed him recreating Majin Buu’s signature hand gesture…
…and Haaland replied: “I mean I don’t disagree.”
The meme has now spread across the internet all over again…
@spartanicartzdxd Haaland looks like Buu #dragonball #haaland #anime #greenscreen
Creators are now optimising for ChatGPT, not just TikTok. And some creators are getting brand deals after appearing in AI answers!
Fashion creator Andrew Polo claims inbound brand inquiries rose 50% after ChatGPT surfaced him for eczema-related searches.
Niche expertise, interviews, press mentions and clear titles all seem to help AI discovery.
YouTube creator content appears in more than 25% of AI assistant answers.
__________
Also this week…
Pepsi deleted a Threads post that said Wild Cherry is what happens when regular cherry “stops asking permission”.
-
Pepsi’s response…?
-
__________
More brand news and views:
How Wimbledon has deliberately worked with creators to shake off its exclusive image
Reddit is cracking down on AI marketing slop with its own AI, as brands plant posts to influence answers from ChatGPT and Gemini
The first YouTube FIFA Creator Cup streams live from Central Park on Sunday
Dentsu is integrating Meta’s Creator Marketplace and Partnership Ads into Dentsu.connect, letting teams handle listening, creator selection and paid activation in one dashboard
"…Whether the Executive Branch (through the SEC) has done enough to hold Mr. Musk to account for his alleged violation is, like many other issues, for our citizenry to decide at the ballot box…"
Over the years, brands have turned community spaces into another place to sell.
Outbound commenting, sponsored comments and SSO tactics are making users more sceptical.
In this guest essay for Rachel Karten’s Link In Bio newsletter, Matthew Stasoff’s (who is awesome) matrix splits community management into service, banter, good times and hero moments.
The strongest play is not more “get that bag sis” replies, but useful customer-first action.
One genuine brand hero moment can be worth 50 desperate banter comments…
__________
Also this week:
As Threads approaches its third birthday, here’s an interview with platform boss Connor Hayes. The platform is less an X clone, and more of a Reddit challenger these days…
Is endless TikTok scrolling stopping you hitting your fitness targets?
WeWard, backed by Venus Williams, lets you lock problematic apps until you hit your step goal.
Its new Walking Mode blocks selected apps until users complete a set number of steps. Someone could lock TikTok or Instagram until they walk 3,000 steps, for example. Goals and blocked apps are fully customisable.
The app rewards walking with “Wards,” which can be exchanged for cash, gift cards or donations.
Facebook users keep accidentally posting onto Threads… and Threads users are loving it.
Once switched on, Facebook’s Threads cross-posting toggle can stay active for future posts
That has led private-feeling updates, family messages and awkward posts into public Threads feeds
Meta says it has added more friction to reduce accidental cross-posting.
-
__________
More WTF this week:
The search for Mark Zuckerberg's mystery stylist is on
OnlyFans creators are accidentally helping to clean up hacked government websites
A Facebook argument is being linked to a fatal family stabbing in Alabama
Discord accidentally banned over 8,000 people for posting grids and other ‘benign’ images.
A safety-system bug wrongly flagged grid-like pictures, including chessboards and Minecraft inventories.
Discord says around 200 users were affected for posting grid-style images.
Another 8,000+ accounts were banned after posting other benign images since May.
The bug also stopped cleared accounts from being automatically unbanned. 😬
__________
More fails spotted:
Donald Trump isn’t as popular on TikTok as he thinks. He claimed he was found to be the number one account on the platform, but he’s actually just the most-watched world leader, a smaller pool of competition!
You could have less control over your TikTok FYP than you think
A Northwestern study found negative feedback works, but often only temporarily.
The “not interested’ button reduced unwanted videos by around 84%. Simply skipping videos cut them by only 48%.
But a short re-engagement with unwanted content can cause the algorithm to “relapse”.
Your FYP is trainable, but it needs constant discipline to stay trained.
$1.4 TRILLION
__________
More social stats:
BAD FOR LADS: LBG Media, behind LadBible, says Meta’s creator-content push helped drive a 41% drop in indirect revenue
FIGHTBACK: Meta responds to the Indian child abuse ad scandal says it removed 4 million+ suspicious accounts and 36 million child exploitation posts last year
AI SPOTTING: TikTok says it has labelled more than 3 billion videos as AI-generated or AI-edited
TOO MUCH: 41% of longform LinkedIn posts were likely fully AI-generated in a new study. On X, around 29% of longform posts were fully AI-written, with 23% AI-assisted
TOP OF THE PODS: 29% of weekly UK podcast listeners now choose YouTube as their main service, making it Brits’ top podcast platform
Here are the top 50 highest-paid creators in 2026, charted…
-
__________
Also spotted this week…
WhatsApp is facing pushback from regulators in India and Somalia over fraud concerns around its username rollout. [TNW]
UK watchdogs advise private profiles, limited sharing, and Close Friends groups for children’s photos as AI-generated child abuse images were up 14% year-on-year in an IWF study. [The Guardian]
The EU is moving closer to a social media ban for teens. [Politico]
The EU’s child abuse scanning bill is back from the dead. [Politico]
Denmark has joined an EU court fight over whether platforms must pay publishers. [TNW]
A French watchdog has ordered Meta to negotiate with publishers in a news copyright case. [Bloomberg $$$]
An app to sync Shopify with Google and YouTube could cause sellers real headaches from 18 August. [Search Engine Roundtable]
New features and tests:
WhatsApp is rolling out new features that make it easier to manage conversations on Wear OS. [WABetaInfo]
WhatsApp is updating the group chat info interface on iOS. [WABetaInfo]
Instagram is testing a slightly tilted to the left UI for story stickers. [@oncescuradu]
Substack expanded its tools for community, podcasts and analytics. [On Substack]
Netflix is adding short-form video from top publishers as it competes with YouTube and TikTok. [TNW]
How ByteDance’s Seedance is quietly becoming Hollywood’s favourite AI video tool…
__________
Also worth a read:
Inside the big business of the creator economy, with the agents making it happen
At 17, she sued Meta and Google, and won. Now she’s ready to tell her story
Influencer screenings aren’t going away
Will late-night TV work on YouTube?
Meta’s Bangkok office was briefly taken over… by a rogue squirrel! 🐿️
The animal reportedly arrived inside a package before escaping into the office.
It spent around 20 minutes running past staff before being caught.
One janitor was scratched on the finger and later taken for a medical check.
At least the incident gave Meta employees a rare moment of comic relief after all the layoffs and morale issues lately!
That's your lot for this week.
In an ironic plot twist… 👇
-
Giphy
Just as i’m about to hit send, and after spending much of this week’s Geekout calling Meta creepy, an email from Meta just landed in my inbox!
They want to gift me a pair of its Ray-Ban Meta AI smartglasses (the Blayzer Optics ones. Fancy).
So, once they arrive, I'll field-test whether they're a genuinely useful bit of tech… or £400+ of expensive creep-stalker hardware strapped to my face.
Bold of Meta to hand the reviewer's job to the bloke who just wrote the "Meta can't stop being creepy" headline, but here we are.
-
[Imag: Giphy]
-
Goodbye, geeks!
-
P.S. This newsletter is edited by Martin SFP Bryant
