The wildest AI art moments of the year

Let's not beat about the bush. 2025 was the year AI stopped pretending to be helpful and started actively competing for our livelihoods. We pretty much went from "AI is just a tool" to "AI is the future of the workforce" in 12 months flat. And if you're a creative professional, you've probably spent most of this year watching the ground shift beneath your feet.

It wasn't the technology itself that shocked us. It was the creeping realisation that most people genuinely don't care whether what they're consuming was made by a human or a machine. 

A fake rock band racks up a million Spotify streams? No problem. Christmas adverts generated by algorithms? Fine, apparently. Social networks devoted to AI slop? Sign me up. The year didn't just normalise AI art; it revealed that for a huge chunk of the population, authenticity was never really the point in the first place.

And nowhere was this more obvious than in the absolute deluge of AI-generated rubbish that flooded every corner of the internet.

Slop that didn't stop

The sheer volume of AI slop in our feeds this year was staggering. New apps like Google Veo 3 made it all possible, allowing anyone to generate nightmarish content that breaks every rule of the physical world.

Particularly excruciating were the AI ASMR videos: people eating molten lava, kneading rocks that defy physics, crushing fired metal like it's plasticine. The aim wasn't relaxation; it was to trigger visceral disgust whilst keeping you watching.

Then there was the AI gastro-horror trend, where the food either ate itself or consumed the people supposedly eating it. David Cronenberg couldn't have dreamed up some of this stuff. And yet huge numbers of people watched, liked and shared it. Not because it was good, but because it was compellingly awful; a digital car crash you couldn't look away from.

The marketing world, bless them, wasn't about to miss out on this craze. McDonald's Netherlands treated us to an AI Christmas ad depicting the festive period as "the most terrible time of the year"; complete with exploding trees, burnt baking and dinner table disasters.

They pulled it quickly after the inevitable backlash, claiming it was "an important learning" about AI use. But filmmaker Jacob Reed had other ideas, using his own AI-generated character to tear into both McDonald's and the ad agency. The satirical response went viral, with the deepfaked character eating a Burger King burger and slagging off McDonald's products. Poetic justice, served cold.

Meanwhile London got its own nightmare before Christmas, courtesy of a development in Kingston-upon-Thames, which hung a massive AI-generated banner featuring grotesquely tortured faces, contorted snowmen, and a dog with a seagull's head. Described as "Lovecraftian horror" by locals, the private investment firm eventually tore it down, but not before everyone wondered how on earth it got approved in the first place.

Deepfake invasion

If 2024 had been the year we worried about AI art, 2025 was the year we started worrying about AI actors. Step forward Tilly Norwood, an AI "actress" who's apparently about to "sign with" a Hollywood talent agency. Her debut was in a parody sketch called "AI Commissioner," which satirised a future where creative decisions are made by machine learning models. Except it wasn't satire at all. It was just... the thing itself.

And then things when properly crazy, as OpenAI launched the Sora iOS app, where you could generate deepfakes of yourself and your mates using the Sora 2 video model. Basically, you make a short video recording to verify your identity, and then the app can insert you into any generated scene. 

OpenAI said this would bring back the "sense of community" that's been lost by other apps. Because nothing says authentic human connection like an entire feed of fake AI content featuring deepfaked versions of people you vaguely know from the internet.

The chaos was immediate. People generated Nazi SpongeBob SquarePants, ads for Epstein Island children's toys, and all manner of copyrighted characters doing things they absolutely shouldn't. OpenAI initially said rightsholders had to opt out if they didn't want their IP used, then quickly backtracked to an opt-in policy after the Motion Picture Association exploded in anger. CEO Sam Altman claimed to be surprised by the controversy, which may go down as the least believable corporate statement in history.

The industry fights back (sort of)

Amidst all of this, there have been glimmers of resistance. Dragon Con ejected an AI art seller from Artist Alley with a police escort, to cheers from other artists. Galaxy Con quickly followed suit with its own ban. It wasn't much, but it was something; proof that creative communities are finally saying "enough."

Meanwhile, pop princess Sabrina Carpenter got roasted for using AI artwork on custom TikTok stickers promoting her album Man's Best Friend. Fans picked apart the wonky features, missing appendages and blotchy text.

Kesha pulled a similar stunt, using AI art for her single DELUSIONAL, then claiming it was meant to echo her frustration with how undervalued artists are. An excuse about as convincing as claiming you stole from a poor box to make a point about capitalism. Both artists eventually replaced their AI covers with human-made alternatives, but the damage was done.

The uncomfortable truth

Here's what ties all these incidents together: they expose a deeply uncomfortable truth about where we are. We're not debating whether AI should replace human creativity any more; that ship has sailed. We're now in the phase where we're discovering that most consumers genuinely don't care who or what made the thing they're enjoying, as long as it's cheap, quick and vaguely entertaining.

The former director of Square Enix put it bluntly: "Gen Z loves AI slop." As evidence, he pointed to Roblox game Steal a Brainrot; a multiplayer game built on AI-generated assets that became one of the biggest games of the year. His argument? The upcoming generation of gamers "was born in" the slop. They don't see AI content as inferior; they see it as normal. Depressing or what?

Netflix's Pokémon Concierge producer went even further, claiming that animators themselves prefer using AI because "it feels inhumane to draw 100,000 pictures all by hand." Which might come as news to, you know, actual animators. Elsewhere Christie's held its first AI art auction and despite fierce opposition from thousands of artists, it exceeded estimates and attracted a wave of collectors. The market has spoken, and it says: authenticity is optional.

Conclusion

So where does that leave us? Probably scrolling through feeds full of content we can't quite trust, made by entities we can't quite identify, whilst the definition of "art" gets stretched so thin you can see right through it. 

So say goodbye to 2025; the year AI stopped being a tool and became the competition. And welcome to 2026, the year when creative professionals will need to fight harder than ever to prove that human-made work still matters.