Italy’s Prime Minister Giorgia Meloni shared a photo of herself on She added: “Whoever made these pictures really made me look better,” but also warned: “The truth is that people these days are willing to do anything to attack and spread lies.” The spread of Meloni’s obscene *deepfake* photo has sparked a new debate: If even a person holding a high position like the Prime Minister is not safe from this kind of manipulation, then what will happen to the photos of ordinary citizens? Let’s delve a little deeper into the world of *deepfakes*…
What is deepfake and how does it work?
The term “*DeepFake*” is a combination of two English words: “*Deep Learning*” and “*DeepFake*.” It is a technology based on Artificial Intelligence (AI), which is capable of creating fake photos, videos or audio recordings, which look absolutely real. There are two main ways to create *deepfakes*:
**Generative Adversarial Networks (GANs):** This process is like a competition between two neural networks. The first – called a “generator” – creates a fake image, while the second – called a “discriminator” – examines the image to decide whether it is real or not. As the generator repeatedly succeeds in deceiving the discriminator, it begins to produce fake images with such accuracy that it becomes almost impossible to distinguish between the real and the fake.
**Deep Autoencoders:** In this approach, the AI system is first trained using hundreds of images of a specific person (e.g. Meloni). Then, it overlays another person’s facial expressions onto that person’s face in such a way that it appears as if the person is making those expressions themselves.
How big is the deepfake* industry?
The market of this technology is growing very rapidly. According to a report by Fortune Business Insights, the global deepfake technology market was valued at US$9.19 billion in 2025 and is projected to reach US$11.18 billion by 2026. By 2034, this market is projected to reach US$51.42 billion, growing at a compound annual growth rate of 21%. However, there are some worrying aspects to this tremendous market growth:
Over the past three years, fraud cases involving deepfakes have increased by 2,137%.
In 2025, 1,567 unique incidents involving deepfakes were recorded, leading to confirmed fraud losses of approximately US$1.3 billion.
By the end of 2026, up to 90% of online content could be created by AI.
How is it possible for your photos to be tampered with?
To create deepfakes, AI needs data. The photos and videos you upload to social media platforms like Facebook, Instagram, and X can serve as input for this process. In Meloni’s case, criminals obtained her photos – and those of many other Italian women – from social media and doctored them. Therefore, any photo or video of you that is publicly available is always at risk of being misused by AI – whether you are a common citizen or a Prime Minister.
So, how can you protect yourself from the threat of deepfakes?
Follow these 5 ways to protect yourself from deepfakes:
Check before you believe: This advice came from Meloni herself. Do not immediately accept any suspicious photo or video as genuine. Use tools: Deepfake detection tools – like Reality Defender, Sensity AI, and Deepware Scanner – can identify images created by AI.
Google’s “About this Image” tool: This feature of Google can tell the real source and history of any particular image.
Digital Watermarking: Countries like Italy and India have enacted laws to watermark content created by AI, so that common people can differentiate between real and fake.
Keep your personal information safe: Avoid uploading private photos on social media. Keep your account set to “Private” and avoid accepting friend requests from people you don’t know. Are there any strict legal measures in place against deepfakes?
Italy has introduced a prison sentence of up to five years for people who cause harm through deepfakes. Meanwhile, India – by amending its IT rules in February 2026 – made labeling of AI-generated content mandatory and set a standard that required removal of illegal content within just three hours.