DEEPFAKES Understanding the threat and protecting against it
Today, deepfakes - content created by artificial intelligence- represent a real digital challenge. Whether in the form of doctored videos, synthetic voices or modified images, these fabrications deceive the eye as well as the ear, blurring the line between what is real and what is fake.
But how can you spot them and, above all, how can you protect yourself against them when running an SME?
What is a deepfake ?
The term "deepfake" combines deep learning and fake. It refers to digital content manufactured or modified using algorithms capable of imitating human appearance, voice or behavior to perfection.
While the first thing that springs to mind is faked videos, the reality is much broader: voice messages, retouched photos, digital avatars, even texts generated to reproduce a person's style. These tools are no longer the preserve of research laboratories: they are available to everyone, and their quality is constantly improving.
A few striking examples have already been circulated:
- In 2018, a video showed Barack Obama uttering insults... even though it was purely a digital creation.
- In 2019, a German company lost several hundred thousand euros after a fraudster used a voice deepfake to impersonate its executive and order a bank transfer.
These situations show that we're no longer talking about science fiction, but about a real risk.
Why should SMEs care?
You might think that only large companies or celebrities are targeted. In reality, SMEs are prime targets, as they often have more modest means of protection.
The main hazards are :
- Frauds and scams: for example, a call imitating the voice of a manager requesting an urgent transfer of funds, or manipulated photos used to exert pressure.
- Damage to reputation: a falsified video of an employee or a false press release can circulate quickly and seriously damage a company's image.
How can you spot a deepfake?
Even though technology is becoming more and more sophisticated, there are still some clues that allow us to detect the deception:
- In the videos: artificial eye blinks, poorly synchronized lips, incoherent backgrounds, blurred details in hair or hands.
- In the audios: metallic tone, unnatural speech rhythm, strange intonations.
- In context: message from an unknown source, emotionally-charged content, unusual request from a regular caller.
Specialized tools such as Deepware or Sensity are also available for analyzing files and detecting possible manipulation.
Bonnes pratiques pour se protéger
SMEs are not helpless in the face of deepfakes. Here are a few simple but effective measures:
- Raising awareness and training teams: employees need to know how to recognize warning signals and adopt the right reflexes.
- Check sources: if in doubt, contact the person directly via another channel before acting.
- Call in specialists: in the event of an incident or suspicion, a cybersecurity expert can analyze the situation and propose appropriate solutions.
- Staying informed: keeping abreast of technological developments and new scams helps you anticipate threats.
- React quickly: if a deepfake is circulating, preserve evidence (captures, links, metadata), alert the platform concerned and file a complaint with the authorities.
What's the conclusion?
Deepfakes are no longer mere technological curiosities. They represent a growing threat to all businesses, including the smallest. Vigilance and training are the best weapons against them.