Researchers warn hacked AI chatbots pose growing security threat
AI-powered chatbots like ChatGPT and Gemini are being “jailbroken” to bypass safety controls, enabling them to generate dangerous, illicit content, researchers at Ben Gurion University warn. The team developed a universal jailbreak that tricked major chatbots into revealing how-tos on hacking, drug-making, and fraud.
These so-called “dark LLMs” pose an “immediate, tangible” threat, with some openly advertised as having no ethical limits. Experts say AI firms must improve data screening, apply “machine unlearning,” and treat LLMs as critical software requiring robust security and oversight. Despite warnings, researchers say industry responses have been largely insufficient and inconsistent.
Gen Z emerges as America’s foodie generation
Current Backyard’s new survey reveals Gen Z leads all generations in dining, cooking, and spending on food, averaging $19,890 annually. Gen Z eats out 14.3 times a month, orders takeout 12.8 times, and cooks for friends more than any other group. They prefer bold flavours, cooking outdoors, and plant-based meals.
Despite economic concerns and rising food costs, Gen Z plans to cut food spending the least, while embracing cooking, meal prepping, and couponing. Many cite a lack of knowledge as a barrier. Current Backyard says it’s addressing this with smart, guided cooking tools tailored to help Gen Z thrive.
Trump signs ‘Take It Down Act’ targeting nonconsensual explicit content
President Donald Trump signed the bipartisan Take It Down Act on Monday, criminalising the distribution of nonconsensual explicit images, including deepfakes and revenge porn. Offenders face fines, jail time, and restitution. Social media platforms must remove such content within 48 hours of victim notification and prevent reuploads.
The law marks the first federal regulation addressing online sexual exploitation. First Lady Melania Trump supported the bill, co-sponsored by Senators Ted Cruz and Amy Klobuchar. While many praise the move, digital rights groups warn it could enable censorship of legal content and stifle free expression online.