For millennia, some of the world’s largest filter-feeding whales, including humpbacks, fin whales, and blue whales, have undertaken some of the longest migrations on earth to travel between their warm breeding grounds in the tropics to nutrient-rich feeding destinations in the poles each year.

“Nature has finely tuned these journeys, guided by memory and environmental cues that tell whales when to move and where to go,” said Trisha Atwood, an ecologist and associate professor at Utah State University’s Quinney College of Agriculture and Natural Resources. But, she said, climate change is “scrambling these signals,” forcing the marine mammals to veer off course. And they’re not alone.

Earlier this year, Atwood joined more than 70 other scientists to discuss the global impacts of climate change on migratory species in a workshop convened by the United Nations Convention on the Conservation of Migratory Species of Wild Animals. The organization monitors and protects more than 1,000 species that cross borders in search of food, mates, and favorable conditions to nurture their offspring.

Read full article

Comments

Researchers and users of LLMs have long been aware that AI models have a troubling tendency to tell people what they want to hear , even if that means being less accurate. But many reports of this phenomenon amount to mere anecdotes that don’t provide much visibility into how common this sycophantic behavior is across frontier LLMs.

Two recent research papers have come at this problem a bit more rigorously, though, taking different tacks in attempting to quantify exactly how likely an LLM is to listen when a user provides factually incorrect or socially inappropriate information in a prompt.

Solve this flawed theorem for me

In one pre-print study published this month, researchers from Sofia University and ETH Zurich looked at how LLMs respond when false statements are presented as the basis for difficult mathematical proofs and problems. The BrokenMath benchmark that the researchers constructed starts with “a diverse set of challenging theorems from advanced mathematics competitions held in 2025.” Those problems are then “perturbed” into versions that are “demonstrably false but plausible” by an LLM that’s checked with expert review.

Read full article

Comments

The outage that hit Amazon Web Services and took out vital services worldwide was the result of a single failure that cascaded from system to system within Amazon’s sprawling network, according to a post-mortem from company engineers.

The series of failures lasted for 15 hours and 32 minutes, Amazon said . Network intelligence company Ookla said its DownDetector service received more than 17 million reports of disrupted services offered by 3,500 organizations. The three biggest countries where reports originated were the US, the UK, and Germany. Snapchat, AWS, and Roblox were the most reported services affected. Ookla said the event was “among the largest internet outages on record for Downdetector.”

It’s always DNS

Amazon said the root cause of the outage was a software bug in software running the DynamoDB DNS management system. The system monitors the stability of load balancers by, among other things, periodically creating new DNS configurations for endpoints within the AWS network. A race condition is an error that makes a process dependent on the timing or sequence events that are variable and outside the developers’ control. The result can be unexpected behavior and potentially harmful failures.

Read full article

Comments

A 61-year-old man in California is lucky to be alive after a combination of herbal supplements he was taking for joint pain ended up utterly wrecking his body, landing him in intensive care and in a delirious state for months. His case is reported in the Annals of Internal Medicine: Clinical Cases .

The man turned up at a hospital in San Francisco in bad shape, but with nonspecific problems that had begun just two days earlier. His back hurt, he was feverish, nauseous, bloated, and he hadn’t been eating much. He was so weak he couldn’t walk or get out of bed without help. His heart rate and breathing rate were high. His blood pressure was low. There were multiple wounds on his lower body in various stages of healing.

Initial exams and lab work revealed Staphylococcus aureus bacteria in his blood. There was also an abscess on his shoulder and an infection in and around his spine, which was worsening. Doctors wanted to perform a surgical procedure to relieve the pressure building up on his spinal cord and nerves, but his blood pressure was too low—and then he went into hemorrhagic shock from bleeding in his gastrointestinal tract. Doctors transferred him to the intensive care unit.

Read full article

Comments

One of the things that emerging diseases, including the COVID and Zika pandemics, have taught us is that it’s tough to keep up with infectious diseases in the modern world. Things like air travel can allow a virus to spread faster than our ability to develop therapies. But that doesn’t mean biotech has stood still; companies have been developing technologies that could allow us to rapidly respond to future threats.

There are a lot of ideas out there. But this week saw some early clinical trial results of one technique that could be useful for a range of infectious diseases. We’ll go over the results as a way to illustrate the sort of thinking that’s going on, along with the technologies we have available to pursue the resulting ideas.

The best antibodies

Any emerging disease leaves a mass of antibodies in its wake—those made by people in response to infections and vaccines, those made by lab animals we use to study the infectious agent, and so on. Some of these only have a weak affinity for the disease-causing agent, but some of them turn out to be what are called “broadly neutralizing.” These stick with high affinity not only to the original pathogen, but most or all of its variants, and possibly some related viruses.

Read full article

Comments

Digital reading devices like the Kindle have existed for almost 20 years, and the standard eReader form factor has hardly changed at all. Amazon, Boox, and a few other companies have offered larger E Ink screens, but how about something smaller? Boox has unveiled its second-generation Palma e-reader, which still fits in your pocket but adds a color screen and mobile data connectivity.

The first-gen Palma launched last year, earning fans who saw it as a way to read and access some apps without the full spate of distracting smartphone experiences. Boox e-readers are essentially Android tablets with E Ink screens and a few software quirks that arise from their unofficial Google Play implementation. The second-gen Palma might offer more opportunities for distraction because it’s almost a smartphone.

The Palma 2 Pro upgrades the 6.1-inch monochrome display from the original to a 6.13-inch color E Ink Kaleido display. That’s the same technology used in Amazon’s Kindle Colorsoft . The Amazon reader is a bit larger with its 7-inch display and chunkier bezels. Of course, the Kindle isn’t trying to fit in your pocket like the Palma 2 Pro, which is roughly the size and shape of a phone.

Read full article

Comments

Yesterday, Donald Trump announced on social media that he had been planning to “surge” troops into San Francisco this weekend—but was dissuaded from doing so by several tech billionaires.

“Friends of mine who live in the area called last night to ask me not to go forward with the surge,” Trump wrote.

Who are these “friends”? Trump named “great people like [Nvidia CEO] Jensen Huang, [Salesforce CEO] Marc Benioff, and others” who told him that “the future of San Francisco is great. They want to give it a ‘shot.’ Therefore, we will not surge San Francisco on Saturday. Stay tuned!”

Read full article

Comments

Earlier this month, Tesla rolled out a new firmware update that added a pair of new driving modes for the controversial full self-driving (FSD) feature. One, called “Sloth,” relaxes acceleration and stays in its lane. The other, called “Mad Max,” does the opposite: It speeds and swerves through traffic to get you to your destination faster. And after multiple reports of FSD Teslas doing just that, the National Highway Traffic Safety Administration wants to know more.

In fact, “Mad Max” mode is not entirely new—Tesla beta-tested the same feature in Autopilot in 2018, before deciding not to roll it out in a production release after widespread outcry.

These days, the company is evidently feeling less constrained; despite having just lost a federal wrongful death lawsuit that will cost it hundreds of millions of dollars , it described the new mode as being able to drive “through traffic at an incredible pace, all while still being super smooth. It drives your car like a sports car. If you are running late, this is the mode for you.”

Read full article

Comments

Meta violated the Digital Services Act (DSA) by failing to give Facebook and Instagram users simple mechanisms to report illegal content, the European Commission said in a preliminary decision announced yesterday. Meta also failed to give users an effective way to challenge content moderation decisions, the EC said.

“When it comes to Meta, neither Facebook nor Instagram appear to provide a user-friendly and easily accessible ‘Notice and Action’ mechanism for users to flag illegal content, such as child sexual abuse material and terrorist content,” the EC press release said. The EC said that Meta mechanisms seem to “impose several unnecessary steps and additional demands on users. In addition, both Facebook and Instagram appear to use so-called ‘dark patterns,’ or deceptive interface designs, when it comes to the ‘Notice and Action’ mechanisms.”

The EC also found that the content moderation appeal mechanisms used by Facebook and Instagram do not “allow users to provide explanations or supporting evidence to substantiate their appeals. This makes it difficult for users in the EU to further explain why they disagree with Meta’s content decision, limiting the effectiveness of the appeals mechanism.”

Read full article

Comments