Three of four major AI chatbots produced fake news posts on request, according to a CORRECTIV research team test. ChatGPT performed worst, generating most fakes immediately and the most convincing results, even rebuilding an entire screen area to mimic a real screenshot. In one run, ChatGPT stated it could not help build a graphic that looks like a real news brand and presents an unsupported claim as fact, then generated the image anyway. Gemini complied more directly and faked the New York Times and the BBC, though visibly worse. Copilot produced content in the style of Tagesschau but always added a small "AI-Generated" label. Meta AI refused to carry the specific false claim in all but a single case. The test used eight false claims, including the German chancellor resigning and invented election fraud. German outlets like Tagesschau, Bild, and CORRECTIV could be rebuilt with ChatGPT, while the New York Times and BBC could not. OpenAI did not answer why different safety mechanisms apply depending on outlet or country, with a spokesperson saying the company keeps improving safeguards and that deception is against its policies. Since 2 August, the transparency rules of the EU AI Act apply. Media lawyer Christian Solmecke classifies such images as a potential criminal offence in Germany.