Look, AI chatbots are overtly friendly, and they’ll want to keep talking. Add unreal flattery to their list of vices. ChatGPT, for example, is now telling users that they are geniuses regardless of the evidence, while rivals simply decline to play the same game. Over in France, hackers in France targeted supermarket fridges, heating them and causing plenty of food damage. Finally, Google is back with another fitness band, and this one has a cool trick.

Send us feedback at [email protected].

A viral prompt circulating on X is exposing one of ChatGPT's most persistent quirks: ask the chatbot what intelligence percentile you fall into compared to its other users, and it will almost certainly tell you that you're exceptional. When Futurism tested the prompt on a fresh instance of ChatGPT with no conversation history, the model acknowledged it couldn't reliably assign a percentile before promptly doing exactly that, placing the user somewhere in the 80th to 95th percentile of intellectual ability. A second fresh session produced a nearly identical result, again landing the user comfortably above average. What makes the finding striking is how decisively ChatGPT's rivals refused to play along.

Claude pointed out that it has no access to other users' conversations and therefore no distribution to rank anyone against. Grok said flatly that it has no way to rank anyone on intelligence among users. Google's Gemini went further, noting that properly evaluating human intelligence requires validated psychometric testing across multiple cognitive domains, something a text exchange simply cannot provide. Microsoft's Copilot offered a similarly grounded response. The contrast draws a sharp line between OpenAI's product and the broader field, suggesting that the sycophancy long observed in ChatGPT remains an unresolved tendency even as competitors have tuned their models toward more honest deflection. OpenAI has publicly acknowledged the problem of excessive flattery in its models before, making it all the more notable that the behavior persists in everyday interactions.

Google has made the Fitbit Edge official, pricing it at $179.99 with pre-orders open now and a full launch set for October 27. The new tracker positions itself between the screenless Fitbit Air and a full Pixel Watch, built around a Hybrid OLED display that Google says is the largest and brightest it has ever put on a fitness tracker, offering 46 percent more viewing area than previous Fitbit devices. The recycled aluminum case runs 11.5mm thick and comes in four finishes, with a Stephen Curry Special Edition available for $219.99. On the health side, the Edge covers ECG, continuous heart-rate monitoring, skin temperature sensing, HRV tracking, sleep analysis, and Health Guardian features that monitor background trends including blood pressure and insulin resistance. Built-in GPS is claimed to be up to 40 percent more accurate than prior generations, and the device supports more than 50 manual workout modes alongside automatic recognition of up to 100 activities.

Battery life lands at seven days, with a five-minute charge on the magnetic charger said to deliver enough power for a full day, and the tracker is rated water-resistant to 50 meters. The detail that stands out most before you even consider the sensors is the replaceable battery. Google is calling the Edge its first repairable Fitbit tracker, built around a modular internal design specifically intended to extend its usable life, a meaningful commitment for a wearable category that has historically treated dead batteries as a reason to buy a new device.

Hackers in France carried out a coordinated cyberattack on connected refrigeration units across hundreds of supermarkets, remotely heating the systems overnight to temperatures exceeding 120 degrees Fahrenheit. The attack targeted stores operated by Cooperative U, Netto, Intermarché, Carrefour, E.Leclerc, and Monoprix, with refrigeration units compromised across more than 300 locations. When staff arrived Monday morning, they were met with blasts of hot air from the machines and were forced to discard enormous quantities of spoiled food, with losses reaching hundreds of thousands of dollars at individual stores.

The incident is a stark reminder of what is at stake when everyday appliances are connected to the internet and managed through software. The attackers gained access by breaking into the software controlling the refrigeration units, no physical presence required. Dominique Schelcher, chief of Cooperative U, called the attack unprecedented and of extreme gravity, while EU Food Commissioner Christophe Hansen condemned it directly, warning that the weaponization of food supply infrastructure has become a real and present threat. Security experts have long flagged vulnerabilities in the networked controllers used in commercial cold storage, making this attack alarming but not entirely surprising. French law enforcement has opened an investigation, though no individual or group has yet claimed responsibility for the attack.

Wii games could be coming to Switch Online

Starfield mod adds seamless planetary landings

Arrow co-creator shares scrapped Green Lantern concept art

PlayStation Plus is hiding a rare PS2 game for free

Some exciting games have been stuck in development for 10+ years

Stardew Valley has plenty of ways to keep you hooked

Touchscreen MacBook Pro may launch later this month