What makes AI-generated "fake news" a new kind of threat compared to older misinformation?

AI can now fabricate photorealistic images and video of events that never happened — "deceptively real" fakes that the eye can't distinguish from genuine footage.

A striking example: a convincing AI-generated image and video of the White House on fire, captioned täuschend echt ("deceptively real"). The danger is a step-change from text-based hoaxes:

• Fabricated visuals carry instinctive credibility — we trust photos and video far more than words, so a fake image bypasses the scepticism we'd apply to a written claim.
• The cost of producing a convincing fake has collapsed — anyone can generate one in minutes, so volume explodes.
• AI-generated content can simply be wrong even when not malicious — models hallucinate details, so a plausible-looking image may depict things that never existed.

This puts the verification burden back on the consumer: you can no longer assume that seeing is believing.

Tip: The reflex to cultivate is "where did this image come from?" before "what does it show?" — provenance now matters more than appearance.