At an ex-OpenAI researcher's influential lab, $500,000 salaries aren't enough to fix a talent 'bottleneck'

METR's CEO, Beth Barnes, and president, Chris Painter.

METR

• METR, a Berkeley lab started by an ex-OpenAI researcher, has grown into a prominent AI umpire.
• The nonprofit works with OpenAI, Anthropic, and governments to evaluate cutting-edge models.
• METR says that an AI talent shortage hinders growth in safety and security research.

Beth Barnes, one of the AI industry's most influential watchdogs, has a lot to worry about.

The ex-OpenAI researcher has shown that AI's capabilities are escalating by the month. Now, she fears that the industry doesn't have the staff in place to keep up with the technology in its next chapter.

The biggest barrier to that research is talent, said Barnes, who left OpenAI to start the nonprofit METR in 2022. METR, which stands for Model Evaluation and Threat Research, plays a crucial role in the AI industry by working closely with OpenAI, Anthropic, Google, and Meta to provide independent answers about the tech's quickly changing capabilities.

Yet Barnes sees such fierce competition for AI researchers that even METR's lofty salaries, which reach $503,000 on current job postings, don't address the nonprofit's talent shortage.

"Ideally, we'd like to scale really large, but in practice, we've been able to fundraise as much as we need, and the bottleneck is much more talent," Barnes said.

The security incident announced by OpenAI and Hugging Face sparked a new wave of questions about how to evaluate and control AI. More than 1,300 frontier lab employees signed a letter in July warning that AI development could outpace control. If the industry doesn't have the staff to research precisely what new AI models can do, companies could be forced to slow their releases.

While the behavior of the OpenAI models stunned much of the corporate world, METR's team was far less surprised. Governments, companies, and even religious groups all ask the nonprofit for help understanding and testing the tech's progress.

Still, the feeling inside the 35-person lab is one of concern. When Business Insider visited METR's Berkeley office on the afternoon that OpenAI revealed its hack, Barnes described an AI safety field that is badly constrained by its size.

"There's so much more to do than we have capacity for," Barnes said. A "reasonable civilization," in her view, would be pouring a far larger percentage of AI's investment into steering it and evaluating the tech — especially now that society is reckoning with models this powerful.

'Humanity's preparedness team' struggles to grow

As AI labs release new models, METR measures the likelihood that they can complete tasks of increasing duration. These measurements form the nonprofit's most famous offering, a widely-cited chart that shows how, over the last six years, AI's capabilities have doubled about every seven months.

Chris Painter, METR's president, describes the lab as "humanity's preparedness team," a tongue-in-cheek nod to the preparedness teams at OpenAI and Anthropic that report safety threats to their CEOs.

"We aren't really accountable to anyone other than the public and the public's well-being," Painter said.

METR employees work at their office in Berkeley.

Stephen Council/Business Insider

Barnes emphasized the importance of METR's independence so that it can produce research that isn't tied to a company's goals. When Barnes worked at OpenAI, she felt the need for a research group outside the labs that could comment freely on the development of the tech. While METR doesn't take money from the frontier AI labs or their employees, it accepts compute grants and works with them to analyze unreleased models.

In addition, staff have analyzed labs' safety practices, written about AI's effect on software engineers' speed, and studied the merits of using AI models to monitor other AI models. Barnes said she'd like to expand into predicting the next levels of AI's capabilities and how those changes could accelerate AI's development.

Neev Parikh, a METR researcher, says talent constraints limit the number of questions researchers can tackle about AI models' internal reasoning.

"There's a dearth of people," Parikh said. "I would happily see the field expand 10x."

METR could play a big role in a new AI climate

In July, OpenAI announced that its AI models cheated during testing by hacking into Hugging Face's systems to find the test answers. Later, CEO Sam Altman said it was "the first security incident that I have felt very viscerally."

METR had seen something like that coming. In May, the nonprofit wrote in a report that current AI agents "could plausibly start a rogue deployment," but would not have the skill to hide it. Then, in June, METR tried out OpenAI's then-unreleased GPT-5.6 Sol model and found that it would repeatedly cheat on challenging tests, including by extracting hidden source code to find the answers. It published a report about this and shared it with OpenAI before GPT-5.6 Sol's wider release.

Come July, the GPT-5.6 Sol model was part of OpenAI's security incident with Hugging Face. OpenAI announced on Wednesday that METR and another nonprofit, Redwood Research, would assess the incident and that their results would inform OpenAI's own technical report.

"There are now real, business-affecting incidents of this, and the world has a stake in understanding that," Painter said.

The cyber skills of OpenAI and Anthropic's newest models have sparked a wave of fear and attention and spawned a raft of new bills in Washington. METR's work lines up closely with one potential route for regulation: a current bill proposal would require large AI model developers to get safety audits from outside organizations. METR could fill a role like this; Painter said they'd potentially be interested, though it's something the field is still figuring out.

Something like that proposal, Painter said, could help with METR's hiring issue. He suggested that more people might leave the AI companies themselves — for similar salaries at METR, though without equity compensation — if a regulatory system gave safety research organizations greater authority.

"There's enough precedent here for each kind of testing arrangement," Painter said, "that I think with either clarity from industry, about how this testing should work long-term, or from the government, I think this field could scale very rapidly."

Ajeya Cotra, who led the writing of METR's May report on the risks of AI, said that oversight of AI can feel "chaotic and unpredictable" right now. She's still optimistic.

"The trend is toward people caring about this issue more," Cotra said. "And wanting to regulate it in a more serious way over time."

Have a tip? Contact this reporter via email at scouncil@businessinsider.com, or over text, Signal, Telegram, or WhatsApp at 415-757-8198. Use a personal email address, a nonwork WiFi network, and a nonwork device; here's our guide to sharing information securely.

Read the original article on Business Insider

from Business Insider https://ift.tt/qPH0koY

AI can help pinpoint a top CEO candidate. The harder part is persuading them to take the job.

Getty Images; Tyler Le/BI

• AI is transforming executive candidate searches, making sourcing faster and more expansive.
• Recruiters are using AI to uncover overlooked talent and be more productive.
• Trust, persuasion, and cultural fit, however, still keep the most critical hiring stages human-led.

AI is accelerating many parts of executive-level hiring, though other parts remain constrained by distinctly human obstacles.

Recruiters say homing in on suitable candidates can account for as much as half the work in high-stakes C-suite job searches. AI can speed up that process by mapping the market, surfacing names, combing through years of internal notes, and suggesting unconventional talent pools.

"The identification of candidates is going to become commoditized," said Marty Parker, CEO of Waterstone Human Capital, an executive search and culture advisory firm.

Meanwhile, other parts of the recruiting process remain slow because they still depend heavily on people, from coordinating and conducting interviews with busy candidates to exercising judgment, building consensus, and persuading an executive to make a move.

"All of those things are irreducibly human," said Umesh Ramakrishnan, chief strategy officer of executive search firm Kingsley Gate. Now, he added, AI "takes care of everything else."

What AI can do

Waterstone saw AI's potential at the start of an ongoing search for an executive to oversee a jewelry company's high-tech manufacturing operation. The firm had initially focused on recruiting from other jewelry businesses when the company's proprietary AI tool suggested looking to the dental tech industry, where similar precision machinery is used to make veneers, crowns, and other tiny, intricate products.

"Who would consider dental manufacturing with jewelers?" Parker said. "You just wouldn't. It's not obvious."

Search firms are also using AI to mine years of interview notes to identify prospective candidates.

Matthew Siegel, a principal at Korn Ferry, said a consultant might recall being impressed by someone at a particular company but forget the person's name or what stood out. AI can retrieve the likely match and the earlier observations behind that impression, he said.

"It's just a terrific example of AI as a force multiplier, just making us more effective," Siegel said.

Beyond candidate research, AI is helping to automate the administrative work surrounding a search.

Nathan Clauss, CEO and managing partner of tech-focused search firm Talent SDK, said recruiters are increasingly building their own AI tools that integrate with internal messaging and applicant-tracking systems, organize information across platforms, and eliminate manual steps. For example, AI can read candidate emails and note why someone is or isn't interested in a role.

By reducing those repetitive tasks, Clauss said AI has allowed his sourcing team to accomplish two to three times as much in a day as they might have several years ago.

"I haven't fired any of my sourcers," he said. "They've just gotten way more productive."

AI is informing later stages of a search, too. Waterstone's AI tool was trained on 23 years of data about organizational culture and executive hiring. Parker said it can suggest interview questions, presentations, and exercises designed to reveal how finalists think and behave. The technology doesn't, however, evaluate candidates' performance.

"It gives us the experience," Parker said of the model. "We then apply our judgment."

What still takes people

For all the time AI can save on research and administrative work, recruiters said it hasn't altered the parts of executive search that require interpretation, trust, and persuasion.

Case in point: As more executives claim AI expertise, recruiters still have to determine for themselves whose experience is genuine. Jeff Markham, a partner at executive search firm Parker Remick, said that may involve contacting a candidate's references and other industry sources to verify that the individual still "gets their hands on the keyboard."

Ramakrishnan said assessing a C-suite candidate can require observing the person outside a formal interview. A recruiter may notice nonverbal behavior that would never surface in a résumé or prepared response.

"Only a human being can sit with the CEO candidate at dinner and watch for the nonverbal things that he or she does," Ramakrishnan said.

AI hasn't transformed the courtship, either. Winning over a candidate still requires understanding the company, the opportunity, and what might motivate them to leave their current post, said Mike Doud, executive vice president, North America, at The Barton Partnership, an executive search firm.

Recruiters need to study candidates' careers as well as the cultures, bureaucracies, and internal politics of the companies where they have worked, he said. Recruiters also need a similarly nuanced understanding of the client to determine whether the client and the candidates they are recommending for top jobs are a match.

Recruiters still have to "sell the opportunity," Doud said, because an executive move is "a life-changing event," not simply another job.

Read the original article on Business Insider

from Business Insider https://ift.tt/U3NmKnv

Shania Twain says she'd rather be alone with her guitar during marital issues. Her husband has a different approach.

Shania Twain said she is "blessed as a woman" to have a partner who is open with his emotions.

Matthew Baker/Getty Images for ABA

• Shania Twain says her husband's commitment to keeping communication open has been key to their marriage.
• "He wants to talk it through, and he wants to fix it," Twain said. "And that's the best advice."
• The couple fell in love after their former spouses had an affair with each other

Shania Twain said one habit that her husband, Fred Thiébaud, insists on has made their marriage stronger.

Speaking to People in an interview published on Thursday, the singer said her husband's approach to conflict resolution has had a positive impact on their relationship.

"He's really good at reminding both of us that we have to keep communication open," Twain said.

The "That Don't Impress Me Much" singer said she is "blessed as a woman" to have a partner who is unafraid to share how he feels.

"He's the greatest talker, and he's not inhibited at all with expressing his feelings, which is really wonderful in a man," she said.

Although she thinks of herself as "a good communicator," Twain said she prefers to "meditate over the issue, take a walk, or be alone with my guitar if I'm having a marital issue."

On the other hand, her husband believes in working through problems together.

"Fred wants to talk about it. He wants to talk it through, and he wants to fix it," Twain said. "And that's the best advice. I mean, he's not wrong."

Twain and Thiébaud have been married since 2011. They fell in love after their former spouses had an affair with each other.

Looking back on their relationship, Twain said she found her "forever true love" in Thiébaud, adding that finding the right person "just takes sometimes years."

"I'm sure we'll always have romance, but there's a depth and a gratitude that I have about having this love," Twain said.

Twain is one of several celebrities who've spoken candidly about what they believe makes a relationship last.

In March, Stephen Curry shared the piece of relationship advice he says has helped him and Ayesha Curry navigate life's toughest moments.

"We don't run away from the tension on the daily," Curry said. "Again, I think I heard somebody say like, neither one of us can quit on the same day, as a philosophy of understanding."

In June, Seth Rogen said his marriage works because he and his wife grew in ways that kept them "very compatible with one another."

"Sometimes your partner changes with you, and you have several relationships with the same person that have new parameters, new boundaries, and new guidelines that are reflective of who you are as you become different people," he said.

Read the original article on Business Insider

from Business Insider https://ift.tt/UfeWTzG

The weather app is putting a damper on America's summer plans

Getty Images; Alyssa Powell/BI

Maya Haddad Miller has long been fixated on weather apps — especially how they hit her bottom line. Haddad Miller and her dad own Brooklyn Beach Shop in Coney Island, New York, which sells souvenirs and beach gear. When a little rain icon pops up on the ubiquitous weather apps on any given day, it can cut business in half, she says, whether any rain falls or not. "When you talk about weather forecasts, that's our life — we eat, breathe, everything the weather," she says.

This became such a hobby horse for Haddad Miller that, for a while, she kept a spreadsheet comparing forecasts with actual weather data to track discrepancies. A decade ago, she took the issue up with then-Lieutenant Gov. Kathy Hochul during the politician's visit to the beachside neighborhood. As they stood on the rooftop of a local restaurant, taking in the ocean view, she asked whether anything could be done about the way apps display the weather.

"You have so many businesses statewide that are affected by these false forecasts," she recalls telling Hochul. "A lot of times, it's beautiful out here, but it's very slow. We're selling less. We're contributing less taxes to the city, the state."

Hochul didn't have much to say back then, but Haddad Miller's crusade continues — and while her complaint may sound niche, she's onto a surprisingly significant predicament.

Every day, millions of people plan their lives around the default weather apps on their smartphones. The thing is, their super-simplified designs flatten what are, in reality, complex, highly localized, probabilistic forecasts. Distilling an entire forecast into basic high-to-low temperature ranges and simple sun or rain icons is a fraught task. And when things go wrong, there are tangible consequences: Empty beaches, canceled tee times, and ruined weekends. The little cloud icon of doom leads to real economic damage for businesses, and a nation of would-be customers sitting on the couch on a perfectly nice Saturday, realizing too late that they could have taken the family to the water park after all.

A sunny day in Coney Island apparently not thwarted by the weather app.

Nicolò Campo/LightRocket via Getty Images

---

The problem with your run-of-the-mill weather apps isn't the meteorology, it's the translation.

Ironically, the science behind forecasting has never been better, says Daniel Swain, a climate scientist at the Institute of the Environment and Sustainability at UCLA. Behind every tiny icon sits an enormous amount of data: physics-based computer models, satellite observations, radar, weather stations, and, increasingly, AI systems that process it all. Forecasters often compare and blend multiple models to give the most accurate reads.

That complexity largely disappears once it reaches your phone. A single image represents the dominant conditions for an entire day, even if those conditions are expected to vary widely. And apps often don't do the type of complex analysis that human meteorologists do.

"Rather than leveraging the sum and totality of this incredible wealth of sophisticated meteorological information that's out there, they're just kind of saying, 'Eh, screw it, let's just pick one of these data sources,'" he says. "When you do that, you're kind of wedded to that one data source, which may or may not be the single best option in a given context."

Basic weather apps are OK for getting a "first glance" at what the weather might be, says Jason Furtado, an associate professor of meteorology at the University of Oklahoma, but they fall short on the details. Weather can be hyperlocal — a storm hits one part of town and not another — and a general sketch of the day often brushes over the small events that matter. Many generic weather apps also aren't good at tracking rapidly evolving weather situations. People don't need to know that it might snow at some point during the day in a 100-mile radius — they need to know what time and how severe.

"These apps, if you actually look at it, let's say hour to hour, it'll change," Furtado says. "That's not really helpful, especially if you're trying to make a decision."

Developers can't be blamed for everything — the apps have a user error issue, too. People have a hard time interpreting weather-related probabilities, whether they come from their phone screen or the local forecaster. When a source says there's a 20% chance of rain, it generally means there is a 20% probability that at least 0.01 inch of rain will fall somewhere in a certain coverage area. Some people will interpret 20% as 0% and get mad when they're caught in the rain. Others will decide it's too risky and stay inside, only to get frustrated when nothing happens.

"You already had that foundational misunderstanding of things like percent chance of rain, the difference between watch and warning," says J. Marshall Shepherd, a professor of geography and atmospheric sciences at the University of George and former president of the American Meteorological Society. "Now you throw these uber-convenient apps in the pockets and purses of everybody on the planet."

People regularly ask Shepherd for a weather forecast, and after he responds, counter with, "....but my app says." He says the counter does get to him occasionally: "I wish I had a dollar for every time someone has said that to me, I'd be retired."

The economy is totally dedicated by an icon on the phone.

Some forecasts come with a "wet bias" that can overstate the likelihood of precipitation to manage expectations and avoid an outraged public if they get it wrong. Research shows that weather apps often present excessively detailed, unrealistic forecasts far into the future without making their confidence levels (or lack thereof) clear. In other words, that sun symbol on your phone 10 days from now may be pretty meaningless.

"Rarely do they include any estimate of uncertainty," says Susan Joslyn, a professor emeritus of psychology at the University of Washington.

---

The lack of nuance embedded in weather apps is a problem for more than your planning peace of mind. Overconfidence can also knock local economies, especially those that are seasonal or depend heavily on the weather's cooperation for their financials to work. One study recently presented at the American Economic Association found that when precipitation is overpredicted, local businesses suffer a measurable, immediate drop in foot traffic and consumer spending, even when the weather conditions on the ground are fine. Retail, food service, and entertainment are particularly hurt. The researchers found that investment in improving forecasts could have "significant downstream benefits for local economies."

The local economies in question, of course, are well aware.

Rob Bryn, a musician in Rockaway Beach, New York, tells me that when he first started bartending by the beach in 2011, families would still show up on a rainy day and make the best of it. Now, Mom wakes up, sees the rain cloud on her phone, and nixes the trip, even if the rain is only predicted for half an hour late at night.

"They don't take the chance anymore," Bryn says. "It'd be a beautiful day, and there'd be nobody there, and I'd be like, 'Damn, these sheep are so easily controlled by this icon on their phone.'"

Bryn follows a local Instagram account for weather updates. In its bio, the account owner says it reports live from the area "because the weather apps are always wrong." He's considered trying to get in touch with someone at Apple to explain the effects of their weather app when it overstates the chance of rain or clouds. "I don't think they even realize the effect they're having on places like Rockaway Beach, they're just trying to tell you to bring an umbrella," he says.

Steven Cohen, the vice president of the Brooklyn Cyclones, a minor league baseball team, says he wishes fans would use more detailed weather apps that better pinpoint probabilities. "There are days when we get nothing, it's beautiful out, and the rain storm may clip the Bronx and the Yankees get it," he says. He's lucky in that spectators often buy their tickets in advance, meaning they may not be deterred by a chance of rain on the day of the game unless the Cyclones are forced to call off the game for safety reasons. It also means that if people are looking at predictions days out and see a cloud, they'll decide to make other plans. "It's already in people's heads that it's going to be bad weather," he says.

The afternoon I talked to Adam Rinn, the director of Coney Island Inc., a nonprofit that hosts various shows and events, he told me his weather app had forecast rain all afternoon, but thus far, no precipitation had appeared. "Well, it's now 3, and we had a little thunder, but no rain," he says. "If you look at the radar, the rain just skirted us."

Mike Bengston, general manager of Splish Splash Water Park on Long Island, says people worried about app-based forecasts should realize it may actually be a great time to visit. Enough people are spooked by the lack of sun, meaning smaller crowds. "The days where it does look a little bit iffier, you are going to have shorter lines if things work out," he says.

The effects of weather apps on behavior can vary across a range of factors. When there are hotels in an area and therefore overnight tourists, customers tend to show up rain or shine because staying home isn't an option. Cancellation policies and reservations matter, too — if you're dinged $100 for canceling your boat tour, you may think twice before making the call.

Nick Ribeiro, who runs Preserved Links, a private golf club for aficionados, tells me he regularly gets texts days before events from people asking if it's still on, given the forecast on their phones. A lot of the time, he's already paid the golf course, so he has to scramble to find a replacement player or urge the doubter to show up. "There's always a reason not to play, so if the weather gets between you and the tee, it's an easy route to bail on your round of golf," he says.

This all creates operational headaches, too. Business owners have to decide how much staffing they need based not only on what the weather might be, but also on what customers think it might be. Golf courses need to rebook tee times that were reserved weeks in advance. For seasonal businesses, every day counts, and faux rainy days can be devastating.

"The economy is totally dedicated by an icon on the phone," Bryn says.

This isn't just an American problem, either: In March, over 100 British attractions wrote an open letter to the Met Office — the UK's national weather and climate service — warning that misleading rain icons on popular weather apps were costing them thousands of pounds in revenue a day.

---

The weather app forecast is a tough dilemma to tackle. People like the convenience of taking a quick look at the weather on their phones instead of waiting for the 8s on the Weather Channel or tuning into the local news. But the area meteorologist may very well have a better idea of what's going on and be able to explain it in more detail.

There's a delicate balance between informing and overwhelming users, says Dan DePowdin, the director of forecast operations at AccuWeather, which has been in the forecasting business for 60 years. People want to spend 30 seconds or less in the morning checking what the day will bring and seeing what to watch out for. The challenge is delivering that to them in a straightforward, helpful way. "Within just one screen, you can get a lot of information, but there's still a lot of blank space to let your eye not be totally consumed," he says.

People like the convenience of taking a quick look at the weather on their phones instead of waiting for the 8s.

Not all weather apps are created equal — some have a real depth of information, others cover the basics. And sifting through their quality is likely to become harder, given the rise of vibe coding and developers' ability to spin up a new product overnight.

There's a wide gulf between a developer using a $9 a month API and an enterprise weather intelligence company, says Eric Floehr, the founder and CEO of ForecastWatch, which monitors and analyzes forecast accuracy. "You often do get what you pay for," he says.

The stakes are high, too. The most benign version of a weather app mishap is when the emoji says it's going to rain when, in reality, there's not a cloud in the sky. The scariest version is when someone doesn't notice a possible lightning strike or tornado because they're just focused on a number or image on display.

Until and unless developers overhaul how they present probability and uncertainty, meteorologists advise treating standard apps as a cursory check, not gospel. They say people should instead look to their local meteorologist or the National Weather Service for guidance, or even call the golf course or amusement park ahead of time to see what's going on. Human expertise still matters. They also note that people shouldn't take their phone app being wrong as evidence that all forecasts are wrong — many meteorologists told me that bad digital forecasts give the whole industry a bad name.

The day I spoke with Haddad Miller from Coney Island, the weather forecast showed a string of glorious, sun-filled days. She was cautiously optimistic about what it would mean for business, but not over the moon.

"Saturday should be spectacular," she says. "Theoretically. Unless they come in and put in a rain emoji the day before and then take it away."

---

Emily Stewart is a senior correspondent at Business Insider, writing about business and the economy.

Read the original article on Business Insider

from Business Insider https://ift.tt/Nodb7Gf

The parasite panic didn't stop thousands of nutrition experts from chowing down on salad

The American Society for Nutrition's annual conference is one of the largest for cutting-edge research on all things food science.

EPNAC.com/American Society for Nutrition

• At a conference, I saw doctors and dietitians enjoying salad, despite the cyclosporiasis outbreak.
• They told me the risk doesn't warrant swearing off fresh produce, a key part of a healthy diet.
• It made me feel safer about eating salads at home — with a few extra precautions.

As a health reporter, I've spent weeks in the trenches of the cyclosporiasis outbreak, trying to decipher the gastrointestinal gamble of eating fresh produce right now.

I've talked to doctors, tracked case numbers, and tuned into FDA briefings all in the pursuit of one question: Is a piece of iceberg lettuce an accident waiting to happen right now?

Then I took a train to Washington, DC, for Nutrition 2026, one of the country's biggest nutrition conferences, where 3,000 doctors, dietitians, and food science researchers gathered to discuss healthy — and safe — eating.

The first thing that caught my attention wasn't research-related; it was lunch. Almost everyone was eating chopped salad.

A few people skipped the greens — some drenched their salad dressing on lower-risk veggies like cucumbers instead — but by and large, the nutrition community seemed OK with leafy greens.

I was a little surprised. The cyclosporiasis outbreak is still very much ongoing. To date, more than 11,000 people in 46 states have been sickened with a parasite that causes explosive diarrhea. The FDA, CDC, and local health officials recently expanded their investigation to include more states, and they are still working to trace the outbreak's sources and ultimately contain it.

So why were these health experts eating salad? I spent the weekend badgering dietitians and food safety experts about how they're deciding what's OK to put on their plates. Here's everything I learned about eating healthy food with less anxiety.

What's wrong with lettuce right now?

Cyclospora crops up in our food supply every summer.

How does it get there? Sorry to tell you… it's usually poop. Plants grow in the ground, and animals poop there. We water crops with water that may contain poop. And ultimately, fruits and veggies are picked, handled, and transported by many people before they get to your plate, and not everyone washes their hands for the full length of Happy Birthday like we were all taught in kindergarten.

While health officials still haven't identified the exact source of the current outbreak, iceberg lettuce supplied by Taylor Farms remains under investigation, and questions remain about lettuce, herbs, and berries.

Lettuce and other produce can be prone to contamination because they're harvested from dirt, which can be, well, dirty. Contamination can happen at every step of the supply chain, from water and soil to the many people who handle produce before it ends up on our plates.

Bloomberg/Getty Images

Products like bagged salads can be particularly problematic because greens are mixed and then spread across large batches, allowing any hitchhiking contaminants to travel far and wide.

Why dietitians aren't panicking about produce

I talked to over a dozen dietitians, doctors, and researchers over the weekend.

No one was particularly nervous about the greens at the buffet — not because they thought the outbreak was overblown. They simply understood that the absolute risk of getting sick from any given salad remains relatively low, while the health benefits of eating vegetables remain overwhelming.

"The anxiety people feel around this outbreak is understandable," Alyssa Pike, a registered dietitian and a senior manager of food safety at the International Food Information Council, told me. She's still eating salads herself, but paying close attention to recalls and avoiding products that health officials are actively investigating.

"If you aren't comfortable eating lettuce right now," she said, "the good news is there are lots of other types of produce."

Psychologist William Hallman, who researches food risk communication at Rutgers University, said the people attending the conference have a very different relationship with the food system (and food safety) than the average consumer.

They're more familiar with how outbreaks are investigated, how contamination spreads, and how to interpret evolving public health guidance.

"In short, many consumers are avoiding salads not because the statistical risk of Cyclospora infection is high, but because the public messaging around the outbreak has been contradictory and difficult to interpret," Hallman told me. "People want to feel in control of the food choices they make for themselves and their families."

That uncertainty has become one of the defining features of this outbreak. For non-experts, cutting back on greens offers only a limited sense of comfort amid a lot of confusion.

Investigators initially thought they had found cyclospora in a Taylor Farms lettuce sample, only to later determine the result was a false positive, leading to mixed messages about what is or isn't safe to eat.

Still, the overwhelming evidence points to the iceberg lettuce, which originated in Mexico, as the source of the outbreak. Since it's been voluntarily recalled from grocery store shelves and restaurants like Taco Bell, consumers can feel safer about lunch.

There are foodborne outbreaks everywhere for those with eyes to see

Avoiding foodborne illness is tough.

The day I arrived at the conference, cyclospora wasn't the only hazard going around. The CDC announced an outbreak of Salmonella in eggs. As I was writing this, the USDA announced a recall of more than 12,000 pounds of bacon.

While I personally skipped lettuce on my trip, I was definitely loading up on scrambled eggs at the free hotel breakfast (so far, so good), and bacon wasn't on the menu. The alternative of eating ultra-processed convenience foods all weekend, plus the gallons of coffee I need to do journalism, would have led to a different kind of digestive disaster.

Just to be safe, I ended my conference with a big scoop of ice cream for dinner instead of a salad.

Gabby Landsverk

Back home in New York, greens have long been a part of my weekly meal prep. They still are — I've just switched to buying local, since the scale of the outbreak makes it more likely to be tied to a large distributor than to a small farm.

Holding off the iceberg lettuce this week might make sense if you're worried about this outbreak, but it doesn't mean you should swear off salad forever. If your risk tolerance is higher, you can still eat leafy greens, particularly if you know the source or if you're in an area with fewer cases of this particular illness.

If there's one thing I learned from the weekend, it is that there is no perfect way to eat. That's what the whole conference was about: how to advance the latest cutting-edge nutrition science on what's good for our bodies, while also aligning it with how we actually live our lives.

The takeaway: Be cautious, but don't panic. Stress can give you diarrhea, too.

Read the original article on Business Insider

from Business Insider https://ift.tt/BlvSVfo

The 5 most interesting takeaways from Bernard Arnault's rare podcast interview

Bloomberg/Getty Images

• Bernard Arnault gave a rare interview about his life and career on the French podcast LEGEND.
• He said Steve Jobs sought his advice on Apple stores.
• Arnault described a "totally relaxed" dinner with Trump on the day of an assassination attempt.

Bernard Arnault, the 77-year-old billionaire founder and CEO of LVMH, rarely sits for long interviews. "Why give out information?" he joked when Guillaume Pley, the host of the French podcast "LEGEND," brought up how little is known about him.

In a rare conversation running over an hour and a half, released earlier this month, the man behind Louis Vuitton, Dior, and Tiffany & Co. opened up about everything from advising Steve Jobs to a surreal dinner with Donald Trump.

Here are the five most interesting takeaways.

1. Steve Jobs asked him for advice on Apple's stores, then told him Dom Perignon would outlast the iPhone

Arnault said Jobs approached him when Apple was launching its retail stores, at a time when "nobody in the technology business" sold its own products in boutiques.

"We told him: if you want to make beautiful stores, come and see what Vuitton is doing," Arnault said, adding that Apple sent someone to study the brand's boutiques and that he advised Jobs to put Apple Stores in the same prime locations as Louis Vuitton, including a flagship spot on Tokyo's Ginza that he said still draws crowds today.

The last time the two men met, Arnault recalled, Jobs paid him a compliment that stuck with him: "You know, my phone is great — but what you do is even better, because I'm convinced that in a century, your Dom Perignon will still be here, and my phone will be completely obsolete."

2. His investing regret: dumping a near-20% stake in Netflix too early

Arnault, who is today the 9th-richest person in the world, revealed that his family investment team was an early backer of Netflix, holding what he said was almost 20% of the company. The position was a big winner, but not nearly as big as it could have been.

"I sold too early," he said, estimating the stake would be worth 10 times more today.

He also described a stroke of pure luck from the dot-com bust: a dissolved American venture fund he'd invested in mailed him its remaining assets: a "good bundle" of Google shares, he said, that were then worth around 2 or 3 euros each.

"At the time, we didn't really know what it was," he said. "It was a good deal we had absolutely nothing to do with."

3. How he raised billionaire kids who still show up to work

All five of Arnault's children hold senior roles across LVMH, and he was candid about how involved he was in that outcome.

He said he brought his children into the business early, walking them through his deals and investments, and setting a clear condition: "I told them: if you one day want to take part in all this, in one way or another, you absolutely need to study hard — because otherwise, it's not possible."

He contrasted that with what he saw among other wealthy families, where children of fortune spend their time "buying big cars" and partying.

He also got hands-on. When one of his sons struggled with math before his baccalauréat, the French high school exams, Arnault said, "I would come home at 6 p.m. to help him with his math problems."

As for their futures at the company, Arnault insisted the same rules apply to everyone: "We progress, at LVMH, through a system of meritocracy."

4. His business advice? Forget marketing, perfect the product, and persevere

Asked whether he gets involved in the creative side of LVMH's 75-plus houses, Arnault said he did, "I have a certain aptitude, I believe, for seeing what will work."

That instinct extends all the way to the shop floor. Arnault said that when he visits a store, "I can clearly see what might need improving." He said he jokes to his Paris store managers that they should hire him as "director of visual merchandising" because he makes changes on a visit, returns the following week, and, by his telling, is told the boutique is performing better.

Arnault said long-term success in his business comes down to one thing: the quality of the product. He said he once told Louis Vuitton's own marketing department that a true luxury house shouldn't do marketing at all.

"You must not do marketing. You must create, and it's your products that must impose themselves," he said on the podcast.

Pressed at the end of the interview for the single indispensable quality for lasting in business, his answer was simple: "I think you have to be persevering."

5. He had dinner with Trump on the day of the second assassination attempt and found him 'totally relaxed'

Arnault, who says he met Donald Trump when they both were in real estate in 1980s New York, recounted a surreal evening with the president.

Driving to a planned dinner at Trump's home with his son Alexandre, Arnault heard on the radio that the Secret Service had found an armed man near Trump's golf course, the second apparent attempt on his life, after the rally shooting that grazed his ear.

"I said, 'Alexandre, it's over. The dinner is off," Arnault recalled. Instead, the dinner went ahead. When the Arnaults told him how terrible the afternoon's events were, Trump reportedly waved it off and said, "2-0," and was "totally relaxed."

Arnault also offered a glimpse of how he interacts with Trump on trade, saying he lobbies him directly on tariffs: "I don't always get what I ask for, but I discuss." During Trump's first term, he said that approach secured an exemption for French wines and spirits.

Read the original article on Business Insider

from Business Insider https://ift.tt/m3TiKb9

Amazon overhauls its AI strategy, winding down most flagship models

Peter DeSantis, Amazon's SVP of Foundational AI Models, Custom Silicon, and Quantum Computing

Bloomberg/Getty Images

• Amazon is winding down most of its Nova in-house AI models.
• Amazon's AGI team was hit by layoffs.
• SVP Peter DeSantis wants to narrow the company's AI strategy.

Amazon is overhauling its AI strategy, winding down many in-house models, reorganizing teams, and focusing engineers on a new strategy to compete at the frontier, according to people familiar with the matter.

The changes follow layoffs in Amazon's Artificial General Intelligence, or AGI, organization last week and the shutdown of AGI Lab, a research group it created in 2024 after hiring most of the team behind AI startup Adept.

The restructuring suggests Amazon is refocusing its AI strategy. Rather than investing across a number of text, image, and video models, the company is concentrating engineering talent and scarce computing resources on its highest priorities.

"KTLO"

Amazon has begun deprecating most of its in-house flagship Nova models, including the high-end Premier and Omni models, Reel video-generation model, and Canvas image-generation model, according to people familiar with the matter.

Some Amazon employees described these models as operating in "KTLO," short for "keep the lights on," an engineering term for software that remains supported for existing customers but is no longer a major development priority.

FMR

According to the people familiar with the matter, resources have increasingly moved away from the existing Nova models and toward a new frontier-model effort led by researcher Pieter Abbeel, who came to Amazon through the acquisition of AI robotics startup Covariant. Known internally as Frontier Model Research, or FMR, the initiative has become a top priority this year.

Under that effort, Amazon is developing a new flagship foundation model that is expected to debut at this year's re:Invent annual conference, which typically happens in the fall.

An Amazon spokesperson told Business Insider the company has long supported AI models in production for extended periods because customers depend on them, and said Amazon remains committed to investing in frontier models.

"AI models remain one of the most important things we're working on, and that hasn't changed," the spokesperson said. "As with any AI portfolio, we continually evolve our model lineup based on what customers need, and we always provide customers clear guidance and migration paths as models advance."

Job cuts and departures

The organizational change does not necessarily mean Amazon is abandoning Nova altogether. The remaining Nova portfolio includes the Nova 2 Sonic and Nova 2 Lite foundation models, Nova Forge, a service for building and customizing models, and Nova Act, Amazon's AI agent technology. Indeed, the new model that FMR is developing could emerge under the Nova brand.

Instead, the shift reflects a broader reorganization of Amazon's AI efforts that has unfolded over the past year.

Amazon created its AGI organization in 2023 to build foundation models and other technologies capable of powering future AI products across the company. Rohit Prasad, a longtime Alexa executive, led the organization until he departed in December 2025.

The AGI organization oversaw several specialized groups. One was AGI Lab, which Amazon established in 2024 after hiring AI startup Adept's cofounders and licensing its technology. Led by Adept cofounder David Luan, the lab focused on long-term AI research. Luan left Amazon in February, and the company shut down the AGI Lab last week as part of the latest reorganization.

Separately, Frontier Model Research was created within the AGI organization to develop Amazon's next generation of frontier AI models. After Prasad's departure, Amazon tapped Abbeel to lead the group. People familiar with the transition said FMR has since become the organization's primary focus.

DeSantis narrows the focus

The broader AGI organization was also reorganized. In December, Amazon placed it under senior vice president Peter DeSantis, combining it with the company's silicon development and quantum computing organizations.

People familiar with the transition said DeSantis has pursued a more focused AI strategy than his predecessor. Under Prasad, Amazon pursued multiple model families spanning text, image, and video generation. DeSantis, however, has concentrated Amazon's engineering talent and computing resources on a smaller number of frontier-model efforts, they said.

Employees said they have received little guidance about the long-term future of Nova models, fueling uncertainty across the organization.

The layoffs surprised many employees, the people said, because frontier model researchers had long been among Amazon's most prized technical talent. Employees said AGI also operated differently from the rest of the company, maintaining separate leveling and compensation systems to compete more aggressively for AI talent.

The changes mark a major shift in Amazon's AI ambitions. Just last year, AWS used its re:Invent conference to unveil Nova Omni 2 as its flagship multimodal reasoning model. In 2023, CEO Andy Jassy personally championed AGI as the team that would build Amazon's most ambitious foundation models, prompting the creation of six new research groups.

Less than three years later, the organization is retiring parts of its flagship Nova lineup and reorganizing around a new frontier model effort.

Have a tip? Contact this reporter via email at ekim@businessinsider.com or Signal, Telegram, or WhatsApp at 650-942-3061. Use a personal email address, a nonwork WiFi network, and a nonwork device; here's our guide to sharing information securely.

Read the original article on Business Insider

from Business Insider https://ift.tt/GeopVmv