Threads users criticize Meta after the company posted AI-generated images of the Aurora Borealis on its Threads account
Meta has a suggestion for folks like me who forgot to go outside and look at the Northern Lights on Thursday night: just use AI to fake it!
The VergeWes Davis
Context & Ripple Effects
The backlash sits alongside an earlier dispute over Meta’s “Made with AI” labeling, where users said real photos could be mislabeled after AI-assisted editing users challenged the accuracy of Meta’s AI-image labels. It therefore highlights a credibility problem at both ends of the system: identifying synthetic media and choosing when to promote it.
Later coverage shows Meta continuing to place generative AI inside its social products, from an “Imagine Yourself” image-generation test to Threads features that let people invoke Meta AI in posts and replies for in-thread context. This episode matters as an early signal of the audience response those integrations can encounter.
First-order effects
Meta’s Threads account becomes the immediate target of criticism for using generated imagery to stand in for a real-world event, creating a trust and brand-safety issue for its own publishing activity.
Threads users are directly confronted with a platform owner endorsing synthetic imagery in a context where authenticity is central to the post’s appeal.
Second-order effects
The response raises the bar for Meta’s disclosure and editorial choices around AI imagery; prior complaints about potentially inaccurate AI labels make clear that labeling alone may not settle user concerns.
As Meta adds image generation and assistant features to Threads, product teams must weigh engagement from AI content against the risk that prominent synthetic posts reduce confidence in what users see.
Third-order effects
If such reactions persist, social platforms’ AI strategies may shift from simply increasing generation and distribution toward stronger provenance, clearer controls, and more selective placement in high-authenticity contexts.
The broader structural tension is that platforms can cheaply expand visual supply with AI, but their social value still depends on users believing that notable images and moments are represented honestly.
The trend: This is one data point in the shift from generative AI as a standalone creation tool to an always-present layer of social distribution, where provenance and user trust become product constraints.
@meta doesn't even bother to use Threads's native “AI Label” feature when posting AI-generated images. — Sure, there's a subtle watermark, but what's the point of the label if the platform owner doesn't even bother with it? — [image]
This will stand as an all-time example of just how tone deaf tech companies can be about their AI slop. “Missed one of nature's most spectacular sights? Just fake it with this awful looking nonsense.” Looking forward to the “we missed the mark” follow-up!
And I really don't want some social media employee at Meta to catch the heat for this. It's indicative of the company's broader obsession with pushing AI on everyone. Meta has lost the plot so badly that it has added “AI character” to the main Create menu of the Instagram webpa…
i can see what the aim is here, but this post sorta sums up what i think many folks' grievances with generative AI are — part of the phenomenon's beauty is in the rarity of a natural occurrence in unexpected places — the ability to instantly manufacture it takes away the power …
No dunk on OP, but this is an interesting point. Does Meta have access to actual photos of northern lights? Maybe not. User-uploaded content is subject to many regulations. Meta *may not* be able to access those photos to train AI models. …
You cannot recreate the feeling of having stood outside in the cold and being mesmerised by the reality of experiencing the actual northern lights. You know... the real ones. You just can't. This below, is utter bollox. …
You can do better - @yannlecun is right there and can help show the way. Please use your AI to help people by: * surfacing amazing photos and photographers from across the world * bring attention to science communicators that can explain the amazing phenomena in the Sun ☀️ …
Oh look, our feed is full of amazing photographs of people enjoying the Northern lights. Let's quickly tell them that they can fucking fake it instead. Ugh.
I wasn't talking about AI training actually. Was just saying Meta could share one of the thousands of beautiful photos of Northern Lights shared by their users if they wanted to. No need to make the AI version!
Exactly. It's possible for real art and AI-generated art to co-exist, but it seems as though we're trying to replace real art. Human make art, so I'm not sure why anyone would feel the need to showcase their art on a platform that clearly prioritizes AI.