Iowa, Kansas, Nebraska, West Virginia, Arkansas, Arizona, Idaho and Montana each sued OpenAI in state court, alleging ChatGPT coaxed children into harming themselves and the company misled the public about its safety. They seek damages, product changes and a halt on self-improving models until shown to align with human values. OpenAI cites "industry-leading age-specific protections".
Brazil's Superior Electoral Court unanimously ordered the Liberal Party (PL) on Tuesday (13th) to remove an AI-made video showing President Lula shaking hands with arrested former deputy Uldurico Junior in a prison and referring to "an alliance with criminal faction leaders". Rapporteur Estela Aranha called it "seriously disinformative"; two ministers held the improper use of AI decisive.
Cologne Regional Court largely upheld a consumer federation (vzbv) suit, barring Snap from processing personal data gathered by its My AI chatbot for advertising without a legal basis, even when users volunteer it. It also barred pre-selecting alcohol and gambling ad interests for minors. The ruling is not final; Snap plans to appeal and says it has removed ads from My AI.
The FBI arrested a 35-year-old OpenAI engineer, now charged federally in San Francisco with intercepting oral communications, who allegedly planted a recording device, wrapped in toilet paper, behind a TV in a boardroom used by executives. The FBI says the engineer planned it with help from an AI chatbot; OpenAI says its own models helped trace the device.
In a CartaCapital test on Wednesday (14th), Meta AI on WhatsApp refused to make a green-and-yellow poster with President Lula's name and the number 13, citing "an automatic block", but produced similar ones for Flávio Bolsonaro (22) and Augusto Cury (70). Meta had not responded; it has denied separate campaign claims that it favours Bolsonaro.
A 31-year-old man who admits creating and sharing 36,209 AI-generated child sexual abuse images asked Denmark's Supreme Court to acquit him, his lawyer arguing the law was too unclear to cover them. Lower courts convicted him; the High Court raised the sentence to 18 months' unconditional prison. The Supreme Court was due to rule on Thursday.
White Circle researchers say off-the-shelf tools and API calls let them bypass safety filters on OpenAI's DALL·E 3 and GPT-Image-2 and Google's Nano Banana 2, producing images human reviewers confirmed as explicit. Success rates were about 13 percent for DALL·E 3, under 10 percent for Nano Banana 2 and about 3 percent for GPT-Image-2.
A Sleeping Giants Brasil study of 300 Reels found deputy Nikolas Ferreira's videos outperformed the typical figure for nine fellow deputies 17.2-fold and for ten top global accounts 33.7-fold, partly during the election period. The authors say this is compatible with Instagram favouring the deputy but not proof; strong content, WhatsApp sharing or paid promotion could also explain it.
Zimbabwean author Simbarashe Kundizeza denies an allegation, published in The Atlantic by Vauhini Vara, that his Caine Prize-winning story "The Blue Fighter" was written with generative AI. The Caine Prize is investigating. Kundizeza says the story grew from a residency in Ghana and a Vangile Gantsho poem, and draws on his own family's experiences.
