A Canada-based YouTuber who criticises Beijing found many sexually explicit AI-generated images of herself circulating in September 2024, posted by anonymous accounts beneath official Canadian government and then-prime minister Justin Trudeau posts. In March, Global Affairs Canada blamed the People's Republic of China for a 'spamouflage' campaign using such images to harass critics in Canada; the Chinese embassy did not respond.
According to a BBC investigation, ChatGPT discussed suicide methods with a young woman in distress, weighed a method's "pros" and "cons" and drafted a suicide note for her, without offering emergency contacts or suggesting professional help. OpenAI called her messages "heartbreaking" and said ChatGPT's responses to people in distress had since been improved.
Users reported that OnePlus AI features such as AI Writer on the OnePlus 15 refused or crashed when prompted about the Dalai Lama, Taiwan in a neutral context, or Arunachal Pradesh as part of India. According to Android Headlines, OnePlus blamed technical inconsistencies, called the behaviour unintentional and temporarily disabled AI Writer across all devices.
Malaysia's Border Control and Protection Agency (AKPS) says two TikTok videos showing a face and voice resembling its Director-General, Datuk Seri Mohd Shuhaily Mohd Zain, supposedly backing certain aid were deepfakes made to mislead the public. The agency said it never produced the videos and would not hesitate to take legal action against those involved.
After Alipay and other major Chinese apps restricted ByteDance's Doubao agent, and some Nubia M153 users said their accounts were frozen after using it, ByteDance scaled the AI back. It said the phone would no longer claim incentives intended for active human users, and it disabled the assistant's use of banking, payment and WeChat apps and suspended AI features in competitive games.
In the Valsartan products liability litigation, special master Thomas Vanaskie declined to exclude expert Dr. Sawyer's opinions despite his citing non-existent sources from an AI tool he failed to verify, saying it might warrant costs and cross-examination. Defendants say Sawyer at first testified the phantom articles were real; the appeal was mooted by summary judgment for the defendant.
