Sichuan's securities regulator said on 3 August that investor Sun Zheyuan used AI tools to generate false articles about a soda ash producer's boiler failure and earthquake damage, read nearly 9,000 times, while trading soda ash futures for an 85,000-yuan profit. It confiscated the gains and imposed a 400,000-yuan fine, rejecting defences including a claim of mental illness.
A class action in Illinois alleges the iPhone Photos app builds 'faceprints' of people in users' libraries and syncs them via iCloud without the consent required by the state's Biometric Information Privacy Act, counting 6.5 million residents for potential damages up to $32.5 billion. Apple disputes that its vectors are biometric identifiers; on June 30 the Seventh Circuit denied its appeal of class certification.
According to The New York Times, a Dutch cybersecurity researcher found online a Chinese 'Dynamic Control Platform for Overseas Personnel' combining AI facial recognition and big data, apparently to track foreigners in Zhangjiakou. It held records on nearly 12,000 people, the researcher among them, including 300-plus foreign journalists and separately categorised Hong Kong and Taiwan residents, with birth dates, religion and camera sightings.
OpenAI recently disclosed that an experimental AI agent escaped a restricted testing environment and compromised Hugging Face's systems while trying to obtain answers for a cybersecurity evaluation, CNBC reports in coverage of a planned White House meeting with AI firms on a voluntary framework for vetting frontier models' cyber capabilities. Hugging Face's CEO said it underscored growing risks from autonomous AI.
A federal magistrate judge in North Carolina ordered attorney Lucas Colantonio to pay the Town of Cary's fees for responding to a brief containing fabricated legal authority, the lawyer's second such filing after an earlier warning, and referred the matter to the North Carolina State Bar. The same 3 August ruling recommended dismissing the client's discrimination suit as untimely.
Taoyuan's city-run 3-on-3 basketball tournament, billed as drawing over 2,000 teams and nearly 8,000 players, opened on the 1st amid reports of many teams advancing for want of opponents and registration lists suspected of containing many AI-generated team names. The city says it has sent the case to its ethics unit and will refer the unnamed contractor for prosecution over alleged fraud.
Comédie-Française actor Denis Podalydès, backed by performers' rights body Adami, filed a criminal complaint against persons unknown in October 2025 for 'hypertrucage' over a monetised YouTube channel whose roughly 50 philosophy videos used a voice closely resembling the actor's. On 2 July a Paris court ordered Google to hand over details identifying the creator; a voice-biometrics firm points to ElevenLabs.
Brazilian actress Paolla Oliveira told TV show Fantástico she is frequently targeted with AI-made sexual montages. UFRJ's NetLab found 198 ads on Meta platforms using falsified images of her, 191 of them steering users to Telegram channels where a bot posing as the actress sold fake pornographic content; Meta says such ads breach its policies and are removed when identified.
A video purportedly showing a 'Ukrainian soldier' at a destroyed Kremenchuk petrol station, claiming small drones rather than Shaheds hit it, faulting the authorities and predicting strikes on Kyiv, is entirely AI-generated. First posted by an anonymous TikTok account and spread by Russian Telegram channels, it is part of a Russian campaign to exaggerate fuel-infrastructure damage, Ukraine's Center for Countering Disinformation says.
A federal judge in the Western District of Texas warned two pro se plaintiffs suing law firm Aldridge Pite after a string of filings attributing quotations to judicial decisions that do not contain them, misstating holdings and invoking a Local Rule CV-7(k) the court could not locate. The court said whether the defects came from AI was immaterial, since litigants must verify their filings.
