Can I still trust images and videos?

A photo used to count as proof. Today, a few seconds of a voice recording are enough to recreate someone's voice, and a handful of photos are enough for a believable video. Criminals exploit this, for example for calls using the voice of a family member, or for faked video calls in which a supposed boss orders a bank transfer.

Technical detection only helps to a limited extent, because fakes and detection tools keep improving against each other. Some providers add invisible watermarks to AI images, and the AI Act requires certain AI content to be labeled. More effective are simple habits: for surprising or upsetting content, check the source first. For demands for money over the phone, hang up and call back yourself. Agree on a code word with your family.

There is a tricky side effect: if everything can be faked, genuine content can also be dismissed as fake.