Social media prank using AI home invader 'bluntly stupid,' police warn (globalnews.ca)

A viral “homeless man prank” uses AI-generated images to convince victims that an unhoused person has entered their home — for example, photos of someone sitting on a couch, lying in a bed, or rummaging through cupboards — sent over TikTok, Instagram and Snapchat. Police from Salem, Mass., to Poole (England) and Ireland’s An Garda Síochána have publicly condemned the trend as “bluntly stupid,” warning it dehumanizes vulnerable people, provokes panic, wastes emergency-response resources, and can create dangerous situations when officers respond to what they believe is an active burglary. The episode underscores broader technical and social risks from easy-to-use generative models and deepfake tools: realistic image synthesis lowers the barrier to producing convincing hoaxes and amplifies harm when content is shared rapidly on social platforms. Authorities’ responses — and parallel controversies such as Zelda Williams receiving AI videos of her late father and the emergence of an AI “actor” prompting industry pushback — highlight urgent needs for better provenance, watermarking/detection tools, platform moderation, public education, and policy coordination to prevent misuse that endangers safety and strains emergency services.