Hacked! – Episode #2
Deepfakes at the 2024 Met Gala: when AI fools millions of people
The Met Gala, a charity event organised each year by Vogue at the Metropolitan Museum of Art in New York, draws the attention of hundreds of millions of people on social media. It is precisely this massive visibility — and the habit of scrutinising every outfit on the red carpet — that allowed images entirely generated by artificial intelligence to fool the world in May 2024, including the mother of one of the victims of this hijacking.
Unlike other cases in our series, this affair involves no malicious financial intent. Yet it illustrates just how far synthetic content, even when created without any intention to harm, can spread confusion on a massive scale.
AI fake images at the 2024 Met Gala: how deepfakes invaded the red carpet
Katy Perry, Rihanna, Selena Gomez: celebrities “present” without being there
On 6 May 2024, as the Met Gala unfolds under the theme “The Garden of Time”, Katy Perry — absent because she is on set — posts two photos of herself on Instagram on the museum steps, dressed in an oversized gown covered in roses. Caption: “I couldn’t make it to the Met, I had to work.”
The problem: these photos do not exist. They were generated by artificial intelligence. Their realism is such that her own mother, Mary Perry, sends her a message congratulating her on her outfit. Katy Perry responds publicly: “Lol mom, you got fooled by AI too, be careful!”
The phenomenon does not stop there. Deepfake images depicting Rihanna and Selena Gomez on the red carpet circulate widely on X (formerly Twitter), even though neither artist attended the event. A post showing a “Rihanna” in an abstract floral gown racks up several million views before a community note flags the content as generated or digitally altered. Some users, confronted with the evidence, continue to insist the images are authentic.
Deepfakes and visual trust: a threat that goes beyond celebrities
From Taylor Swift to the Met Gala: the escalation of synthetic content in 2024
The 2024 Met Gala case comes just four months after a far more serious episode: in January 2024, a fake AI-generated pornographic image of Taylor Swift had been viewed 47 million times on X before being removed.
These two cases, very different in nature, share the same conclusion: it is becoming increasingly difficult to distinguish authentic content from a fabrication, even for a familiar eye — such as a mother looking at a photo of her own daughter. If artificial intelligence can fool a celebrity’s immediate family, it can just as easily produce a false statement, a fake press release or a fabricated public appearance for any organisation or public figure.
Our “Hacked” series documents this same dynamic applied to more directly malicious ends, as in the case of Jeff Bezos, targeted by a wave of disinformation.
Content certification: the answer to the erosion of digital authenticity
Without a proof of authenticity established at source, any visual content can claim to represent anyone, anywhere.
Content certification addresses precisely this issue: it ensures that an authentic image or publication remains verifiable and clearly distinct from any subsequent fabrication — whether malicious or simply viral.
Discover how Certiphy.io protects the authenticity of visual content →
📲 See the original post on our Instagram account: Hacked! Episode #2 – Katy Perry & Rihanna @MetGala
