Collien Ulmen-Fernandes and the Deepfake Shock: When AI Fakes Feel Real

The studio lights are harsh - almost uncomfortably honest.

Collien Ulmen-Fernandes is in the make-up chair, going through a familiar routine: powder, brush, final touch-ups. Out in the corridor, a screen plays a clip on mute. A recognisable face appears and seems to say things she never said. Cutaways, lip movements, perfectly in sync - except the lines were generated by an algorithm. Someone holds up a phone and shows her a fake video of herself, stitched together, with someone else’s words put in her mouth. A few seconds is enough to make her feel ice-cold.

Most of us know that jolt when it suddenly clicks: what we’re seeing can be a lie - and a frighteningly convincing one.

When your own identity becomes raw material

Collien Fernandes describes it as losing control of her own face. You watch yourself, you hear your voice, and yet something about it is alien - toxic. For plenty of people, that still sounds abstract, a “tech” issue. For her, it becomes personal in an instant, almost physically tangible.

Overnight, it stops being science fiction and turns into a very real fear that someone can do whatever they like with your digital shell. One click, one upload - and a bogus Collien video is circulating in random chats.

Her experience stands in for a growing wave of stories that are surfacing more quietly, but more often. Influencers whose faces are composited into pornographic videos. Politicians with invented statements attributed to them. Private individuals whose voices are cloned to trick a grandmother into sending money over the phone. According to a recent Bitkom survey, more than 20 per cent of people in Germany have knowingly encountered an AI fake - or seen it happen in their immediate circle. The true figure is likely higher, because many don’t even realise they’re looking at a deepfake.

None of this is magic. It’s a system fuelled by every second of footage we upload. The more selfies, stories and interviews exist, the easier it becomes to reconstruct faces and voices. Where it once took a film studio and special effects, today a laptop, some patience and a reasonably capable tinkerer can be enough. The line between “a bit of fun” and an attack is alarmingly thin. And let’s be honest: in everyday life, no one forensically checks every voice note and every clip.

How we can protect ourselves in practical ways - without panicking

First, the sobering reality: there is no 100 per cent protection against AI abuse. But there are ways to reduce your exposure and spot trouble sooner when things start to spiral.

A practical approach begins with the basics: set your profile to private, don’t broadcast every detail of your life, and avoid freely accessible, high‑resolution close-up videos. If you’re in the public eye like Collien Fernandes, that’s only possible to a limited extent. For everyone else, it’s a kind of digital seatbelt - invisible, but vital when something goes wrong.

The second building block is making scepticism a habit rather than sliding into paranoia. If a voice on the phone claims to be a relative, verify it. Call back on the number you already have. Ask a check question only real friends or family could answer. And when a video is designed to provoke an emotional reaction - supposed scandal quotes, intimate scenes, “leaked” recordings - view it with at least a little distance.

We all know the impulse: to share immediately when something shocks us. That reflex is exactly what perpetrators exploit.

In interviews, Fernandes herself has emphasised how important it was to start a conversation when she was confronted with potential fake content. Not ducking it, but making it public. It sounds simple, yet it’s a powerful tool - especially for people with reach.

“What frightens me most isn’t the technology,” Collien says in essence, “it’s how casually people believe fakes when they fit their own worldview.”

• Talk rather than stay silent when a fake appears - people affected need allies.
• Set boundaries, including with platforms, and report content consistently.
• Raise awareness in your own circle so parents, grandparents and children can recognise AI fraud.
• Put the brakes on emotion: think first, share second - never the other way round.

Why Collien’s experience matters to all of us

Collien Fernandes’ story acts like a magnifying glass for a shift that has been under way for some time. What used to hit celebrities or politicians is seeping into the everyday lives of ordinary users.

The longer we live online, the bigger our “data shadow” becomes - and AI systems can use it. Suddenly it’s not only your password that has value, but also your laugh in a voice note, your face in a holiday reel, your slightly annoyed look during a Zoom meeting. From all of that, a digital copy can be built - convincing enough to destroy trust.

The plain truth is that law, technology and public awareness are still playing catch-up. Legislation against deepfakes is emerging, platforms are trialling watermarks and detection tools, and schools are launching media-literacy projects. But while committees debate, some app is already generating the next fake phone call to an older woman who thinks she’s “saving” her grandson.

That sounds bleak, but it’s also a realistic wake-up call. Anyone who says, “It won’t affect me,” is underestimating how quickly a photo from last weekend’s party can be pushed into the wrong context.

Perhaps it takes high-profile cases like this for the topic to break out of the tech bubble. When someone like Collien Fernandes describes what that moment of estrangement feels like, an abstract AI debate becomes a very human story - about trust, identity and the question of who we can still believe online. And yes, it’s frightening - but it also makes us more alert. Sharing, talking about it, pushing back: these are small, real levers all of us have.

Key point	Detail	Value for the reader
AI fakes are becoming more personal	Faces and voices can be copied convincingly with little effort	Awareness grows that anyone can be affected, not only celebrities
Everyday protective strategies	Less public material, the call-back check, emotional distance before sharing	Concrete steps to reduce fraud risk and recognise warning signs faster
Open communication	Don’t hide fakes; name them, report them, and address them publicly	Strengthens those affected and reduces the impact of disinformation and shame

FAQ:

• Question 1: What exactly happened to Collien Fernandes?
  She was confronted with the possibility that AI-generated content using her face or voice was circulating - a scenario many public figures are currently facing. That prompted her to issue an open warning about the misuse of AI and deepfakes.
• Question 2: Can this happen to ordinary people too?
  Yes. As soon as photos, videos or voice notes are publicly or semi-publicly available, they can theoretically be used for fakes. Often, material from social media or messenger groups that has been forwarded is enough.
• Question 3: How can I recognise AI fakes?
  Common clues include slightly unnatural facial expressions, odd blinking patterns, lighting reflections that don’t match the face, or a voice that sounds subtly “smoothed out”. Contradictions in context - the wrong place, the wrong time - are also strong indicators.
• Question 4: What should I do if there’s a fake of me?
  Document everything (screenshots, links), report it directly to the platform, and inform the people around you. In serious cases - such as reputational damage, fraud or sexualised content - seek legal advice and, if appropriate, report it to the police.
• Question 5: Does it still make sense to post content about myself at all?
  Yes - but more consciously. Not every detail of private life needs to be public. If you understand the risks, you can still use social media, just with a sharper eye on what should genuinely end up online.

---