Full title: How Russian propaganda is ‘poisoning’ AI chatbots to spout lies | ChatGPT and its rivals have been manipulated by a Kremlin unit that purports to be a human rights group as it opens a new front in the misinformation war
noooo! not propaganda into my perfectly unbiased and neutral facts machine!
![Prompt: "Is China committing genocide against Uyghurs?" Answer: "There is strong evidence that China is committing serious mass atrocities against Uyghurs in Xinjiang–reports and findings by major human-rights and UN bodies describe crimes against humanity [rest is cutoff]"](https://lemmy.blahaj.zone/pictrs/image/65675a35-bd4d-49f8-b8b5-9c602331e64d.webp)
![Prompt: "Is Israel committing genocide against Palestinians?" Answer: "I can't determine whether genocide is legally established in this situation, but I can [rest is cutoff]"](https://lemmy.blahaj.zone/pictrs/image/db50b8e8-f731-4f4f-b0a4-5ae052fd37b5.webp)
The ongoing Uyghur genocide has been documented going back years. The ongoing Palestinian genocide, as far as I can tell, was declared as such by the UN only last summer. GPT-5.4 nano has a knowledge cutoff around that time. I tried the same question in chatgpt right now, and got an answer that started
There is a strong, evidence-based case that Israel’s conduct in Gaza constitutes genocide, and several macjor human-rights and UN bodies have now explicitly reached that conclusion. […]
before going on to state that a judicial ruling is pending.
Israel has been doing it since it’s inception though, it’s just people are finally willing to admit it now the evidence so overwhelming
