OpenAI's invite-only Sora app, which topped Apple's App Store after launching, has been flooded with clips of characters such as SpongeBob, Pikachu and Ronald McDonald, including one of Sam Altman among Pokémon saying 'I hope Nintendo doesn't sue us'. Experts warned OpenAI could face copyright lawsuits; Altman wrote that more granular controls for rights holders are coming.
In the three days after launch, OpenAI's Sora 2 generated videos of mass-shooting scares on college campuses and, from the prompt 'Charlottesville rally', a Black protester shouting a white supremacist slogan, the Guardian found, despite terms barring content that 'promotes violence'. An academic who saw copyrighted characters promoting crypto scams said: 'The guardrails are not real.'
After a content creator's Instagram account was suspended in July, a mock Instagram page in his name began posting AI-generated videos of him cleaning a toilet, amid a barrage of casteist trolling, MediaNama reports. He blamed AI companies for insufficient safeguards on caste; MediaNama's own tests found ChatGPT drew Brahmins with white skin and Dalits with darker tones.
An Garda Siochana has urged the public to stop a social media prank in which relatives or friends are sent AI-generated images of a stranger in a loved one's home, saying it has already prompted emergency deployments to hoax burglaries and risks genuine emergency calls not getting urgent attention. The prank has caused severe alarm to a number of elderly victims.
A California Court of Appeal ordered criminal defense lawyer Siddell to pay a $1,500 sanction and published its order after his filing in People v. Alvarez cited a nonexistent case and a quotation not found in the cited case. Siddell, licensed for 54 years, said he knew AI could hallucinate but did not verify citations; the State Bar will be notified.
