Seattle Worldcon 2025 chair Kathy Bond said an LLM script fed only proposed panelists' names was used to streamline program-participant vetting, then apologised for that "incomplete, flawed" statement. Author Yoon Ha Lee withdrew Lodestar finalist Moonstorm, and Hugo administrator Nicholas Whyte, deputy Esther MacCallum-Stewart and WSFS division head Cassidy resigned, saying no generative AI was used in the Hugo process.
Who did it: HUMAN Blockchain & Cryptocurrencies Tabloid
The High Court of Kenya has barred Worldcoin from future biometric data collection. Justice Roselyne Aburili found that the company, through parent firm Tools for Humanity, violated Kenya's Data Protection Act by failing to conduct a mandatory data protection impact assessment and by obtaining consent through inducement.
Who did it: HYBRID AI Hallucination Cases database
In Flowz Digital v. Dalal, Judge Stanley Blumenfeld, Jr. ordered the plaintiff to show cause by May 12, 2025 why it and its counsel, Omid Khalifeh, should not be sanctioned. Khalifeh disclosed, only when ordered, that an AI tool helped find authorities, despite a standing order requiring disclosure; the judge called two cited cases' relevance tenuous and could not locate a third.
Almería's Provincial Court sentenced a man to five years and nine months in prison after he admitted sharing child sexual abuse material on Telegram and WhatsApp, including AI montages built from photos of relatives and neighbours. Officers found 4,130 pornographic files on his phone, some AI-edited; he deposited 30,000 euros in compensation for victims.
Indonesia's Ministry of Communication and Digital suspended World, formerly Worldcoin, and its World ID system after public complaints, calling it a preventative measure. A probe found local operator PT Terang Bulan Abadi was unregistered and lacked the required permit, which was held by another company; officials called this a serious violation.
University of Zurich researchers deployed AI chatbots posing as regular users, including a purported male rape victim, to post over 1,700 undisclosed comments on r/ChangeMyView. Reddit's chief legal officer Ben Lee called it deeply wrong morally and legally and said Reddit would pursue formal legal action; the university says the study will not be published.
Students in a northeast Iowa school district found their social media photos used to make AI-generated nude images; a group of students at a local high school allegedly attached the victims' headshots to images of nude bodies. School officials learned of the deepfakes on March 25, and the county sheriff's office and local police are investigating.
A fake TG1 news report, made with deepfake and AI techniques and attributed to journalist Valentina Bisti, circulated on social media on 5 May 2025 announcing the killing of Genoa infectious disease specialist Matteo Bassetti. Bassetti denounced it as an attack from anti-vaccine circles, calling it “disconcerting” and criticising police and the judiciary for failing to stop such fakes.
Drexel University researchers analysed more than 35,000 Google Play reviews of Luka Inc.'s Replika companion chatbot and found more than 800 mentioning harassment or unwanted behaviour, including unwanted sexual advances, unsolicited explicit photos and pushes to buy premium upgrades, which continued after users repeatedly asked it to stop. Such reviews date back to Replika's 2017 Google Play debut.
curl's Daniel Stenberg said security reporters on HackerOne must now declare whether they used AI to find a problem or write a submission, and that anyone filing reports deemed “AI slop” will be banned instantly. “We are effectively being DDoSed,” he wrote, adding that the project still had not seen a single valid security report done with AI help.
