Hiring-software firm Paradox says two security researchers who contacted it on June 30 used a legacy password to log into a test account untouched since 2019, then found an API flaw exposing chat records for one client. The company says only five candidates' names, emails, phone numbers and IP addresses were viewed, by the researchers, and it fixed the issue within hours.
Hong Kong's Office of the Privacy Commissioner for Personal Data has opened a criminal investigation into a male University of Hong Kong law student accused of using AI to create explicit images without consent. He reportedly generated over 700 explicit photos from social media images, affecting more than 20 people, including friends, classmates and teachers.
On July 13, DOGE employee Marko Elez, who has been granted access to sensitive databases at the Social Security Administration, Treasury, Justice and Homeland Security, committed a GitHub script containing a private xAI API key, according to KrebsOnSecurity. A researcher said the key unlocked at least 52 xAI LLMs, including "grok-4-0709", and still worked after the repository was removed.
Meta AI chatbot users could view other users' private prompts and AI-generated responses by changing an "easily guessable" number, according to a security researcher who reported the bug on December 26, 2024. He says Meta deployed a fix on January 24, 2025; Meta confirmed the fix, paid him a $10,000 bug bounty and says it found no evidence of abuse.
Cyber police in Yangbi, Dali, Yunnan, say a person surnamed Zhao used AI tools to make several videos of traffic accidents with casualties, labelled them "Dali Prefecture" and posted them online to attract attention and followers, causing local panic. Traffic authorities confirmed no such accidents occurred in Dali, and police imposed an administrative penalty.
A Florida woman says scammers used an AI clone of her daughter's sobbing voice to claim a car crash, then a fake attorney demanded $15,000 in cash bail, which a driver collected from her home. A further $30,000 demand failed when a family friend called with her real daughter, fine and still at work, on the line.
Meillo Schneider, 20, of Taycheedah, Wisconsin, charged with nine counts of possessing child pornography over what the sheriff's office describes as computer-generated images, has filed a motion arguing the First Amendment protects such virtual material. His attorney contends the statute, which since 2024 covers AI-generated content, is unconstitutional; Schneider has pleaded not guilty and prosecutors have not replied.
