Deepfakes and AI-Generated Images

The Pattern

← Safety Recovered Archive / Safety AI can now generate a photo or video of a real person doing or saying something they never did. The result can look like evidence. It is not. For survivors of deception, this matters because a former group, partner, or network can use fake media to punish, silence, or re-trap you. You do not need to become a tech expert to protect yourself. You need to know what is possible, what to look for, and how to slow down before reacting. What it is “Deepfake” usually refers to AI-generated video or audio that replaces a real person’s face or voice. The same technology can also produce a still image from a text prompt, such as a fake photo of someone in a compromising situation. These tools are now widely available and require little technical skill. The image or video may be entirely invented, or it may start with a real photo and alter it. Either way, the person in it did not do what the media appears to show. How it is used against survivors ▌ Fake evidence. A manipulated image or video is presented as proof of cheating, lying, crime, or betrayal. ▌ Reputation destruction. Intimate or compromising fakes are shared online or sent to employers, family, or friends. ▌ Blackmail. A person is threatened with release unless they return, stay silent, or send money. ▌ Reinforcement of deception. A group or person uses fake media to “prove” that critics are evil, sinful, or dangerous. Warning signs ▌ You are shown a damaging image or video with no surrounding context, source, or chain of custody. ▌ The person sharing it demands an immediate reaction and refuses to let you verify it independently. ▌ The lighting, shadows, or facial movement looks slightly off — though the best fakes may not show obvious flaws. ▌ The timing is too perfect: the media appears right when someone wants to control a decision or silence you. What you can do now ▌ Do not share it. Even to defend yourself, sharing spreads the fake and makes it harder to remove. ▌ Preserve evidence. Take screenshots of where it was posted, who shared it, and any accompanying messages. Save the original URL if you can do so safely. ▌ Reverse image search. Tools like Google Images, TinEye, or Yandex can show if the image existed elsewhere first. ▌ Check for metadata. Some images carry metadata showing they were created by an AI tool. This is not always present, but it can help. ▌ Report the platform. Most platforms have policies against non-consensual intimate imagery and manipulated media. Use the reporting flow and keep a record of your report. ▌ Ask a trusted outsider. Someone outside the controlling situation can help you verify and decide what to do next. When to get outside help If intimate images are being shared or threatened, you may have legal options. Laws vary by location, but many jurisdictions now have specific protections against non-consensual intimate imagery and AI-generated sexual material. A domestic violence advocate, legal aid clinic, or cyber civil rights group can help you understand your options. Browse crisis and legal resources → If you know a resource, law, or tool that should be added here, use contact .