25+ min ago (63+ words) [Techmeme permalink] Matthew Hutson / IEEE Spectrum: Researchers unveil PropensityBench, a benchmark showing how stressors like shorter deadlines increase misbehavior in agentic AI models during task completion" " Shortened deadlines and other stressors caused misbehavior" " Several recent studies have shown that artificial-intelligence " This is a Techmeme archive page. It shows how the site appeared at 6:05 AM ET, November 30, 2025....

34+ min ago (603+ words) 45% of mums and dads use AI to help with their children's homework MODERN mums and dads are becoming more and more open to using AI to assist with their parenting decisions, according to research. A report found three in 10 parents of under 18s who use AI tools have done so to give advice to their children, while 33 per cent have considered doing so. Help with homework or school projects is when AI is most commonly utilised (45 per cent), while others turn to it when they want age-appropriate entertainment or books (39 per cent). However, as many as 40 per cent will look to the likes of Copilot or ChatGPT if they need to talk about difficult issues such as bullying, grief or relationships. With 39 per cent of the parents in the study of 5,000 adults finding AI useful to help explain more complex topics…...

34+ min ago (629+ words) 45% of mums and dads use AI to help with their children's homework MODERN mums and dads are becoming more and more open to using AI to assist with their parenting decisions, according to research. A report found three in 10 parents of under 18s who use AI tools have done so to give advice to their children, while 33 per cent have considered doing so. Help with homework or school projects is when AI is most commonly utilised (45 per cent), while others turn to it when they want age-appropriate entertainment or books (39 per cent). However, as many as 40 per cent will look to the likes of Copilot or ChatGPT if they need to talk about difficult issues such as bullying, grief or relationships. With 39 per cent of the parents in the study of 5,000 adults finding AI useful to help explain more complex topics…...

2+ hour, 56+ min ago (302+ words) Google has issued a warning that might make you rethink that "free VPN" you casually downloaded last week. According to the company, some apps pretending to be VPN services are actually dangerous malware wearing a friendly disguise. And instead of hiding your online activity, they may be quietly stealing your most sensitive information. And we're not talking about harmless data either. Google's security team says these fake VPNs can steal banking details, crypto wallet info, passwords, private messages " basically anything valuable stored on your phone. Some can even enable remote access, giving attackers direct entry into your device without you ever realizing something is wrong. The scarier part is how convincing these apps look. Many are promoted through misleading ads or fear-based marketing ("Protect your phone from hackers NOW!"). They rely on the fact that most users don't dig too…...

4+ hour, 24+ min ago (608+ words) Updated Nov. 30 with new warnings on VPN usage and the latest porn ban to come into affect, prompting another surge in VPN installs in the U.S. The seal has been broken. VPNs have operated largely under the radar for years, but now governments are taking an interest. These anonymity masks promise to hide your online activity and access restricted content. They're unlikely to survive unscathed. "Attacks on VPNs are attacks on digital privacy and digital freedom," EFF warns. "And that battle is being fought by people who clearly have no idea how any of this technology actually works." They're not wrong. This is now a disaster waiting to happen. While we have seen plenty of VPN warnings over the years " these have been contained and focused on personal user security. Do not install free apps or Chinese apps neatly sums up…...

4+ hour, 30+ min ago (237+ words) Sunday, 30 Nov 2025 LONDON: The increasing automation and connectivity of cars is not only taking the driving out of driving " and for some, the fun too " but could also be leaving owners vulnerable to hackers trying to remotely hijack their vehicles, analysts say. Cyberattacks on cars are set to become more likely due to the expanding role of software and autonomous driving, according to BMI, a British research subsidiary of financial company Fitch Solutions. The analysts warn that cybersecurity risks to drivers will increase amid "a continued acceleration towards software-defined vehicles," which are being integrated into the broader digital infrastructure. And while the idea is that car companies provide "advanced navigation systems, seamless over-the-air updates and enhanced efficiency and driving experience," the downside is that the vehicles are as exposed to cybercriminals as any other connected device. BMI noted that researchers…...

5+ hour, 35+ min ago (266+ words) A major fire in Gunz top village, Ramban district, destroyed three houses early Sunday. Though there were no casualties, six families were left homeless as their wooden homes were reduced to ashes. The fire, attributed to a short-circuit, was extinguished by volunteers with no belongings salvaged. Devastating Blaze Leaves Six Families Homeless in Ramban A major fire in Gunz top village, Ramban district, destroyed three houses early Sunday. Though there were no casualties, six families were left homeless as their wooden homes were reduced to ashes. The fire, attributed to a short-circuit, was extinguished by volunteers with no belongings salvaged. A major blaze engulfed three houses in a remote hilltop village in Ramban district early Sunday, leaving six families without homes. The fire, originating before dawn in Gunz top village of Trigam, Khari tehsil, spread rapidly to two adjacent wooden…...

5+ hour, 51+ min ago (474+ words) Violent incidents targeting crypto users are increasing, according to warnings shared through industry updates. Recent cases show attackers tracking routines, using social exposure, and targeting individuals with digital assets." Security professionals note that criminals focus on predictable behavior and weak personal safety habits. The trend places new attention on crypto security tips that help reduce everyday risks. The rise in targeted crimes has pushed crypto users to adopt stronger habits, with guidance circulating widely across social channels. A detailed post from a16z's Head of Security outlines how attackers build profiles by observing patterns and collecting exposed details." The post describes incidents involving founders, engineers, and investors who faced kidnappings, forced access attempts, and planned assaults. The examples highlight how attackers rely on casual mistakes, public routines, and oversharing. Daily awareness remains a core theme in these crypto security tips." The…...

6+ hour, 27+ min ago (1063+ words) Pangram detects Gemini 3! Learn more Learn more about our organizational solutions. We made all the results publicly available on iclr.pangram.com. Well, for one, we were offered a bounty! Graham Neubig's tweet offering a bounty for analyzing ICLR submissions In all seriousness, many ICLR authors and reviewers have been noticing some cases of blantant AI-related scientific misconduct, such as an LLM-generated paper with completely hallucinated references, and many authors claiming to receive completely AI-generated reviews. One author even reported that a reviewer asked 40 AI-generated questions in their peer review! We wanted to measure the scale of this problem at large: are these examples of bad behavior one-off incidents, or are they indicative of a larger pattern at work? That's why we took Graham up on his offer! ICLR has a very clear and descriptive policy on what is allowed…...

6+ hour, 30+ min ago (294+ words) Recently I heard a talking head on a finance channel waxing poetic about how wonderful A.I. (artificial intelligence) is going to be for everyone.White-collar workers like herself will be able to focus on important matters while A.I. handles dreary time-consuming chores. Let me present a metaphor, not involving anything technical, that shows how anything can be made to sound wonderful if you are focused on only a single slice of the pie. Of course, the slightest deviation from the core goal of weedlessness reveals the truth: There would be no neighbors, no crops, no nothing. As silly as this metaphor seems, it applies to A.I.The ratio of desirable outcome to total disaster is the same. A.I. "learns" by storing and reviewing alternative choices, factoring in their likelihood of success for the mission assigned.Computers have always been this way, and A.I. is a…...