A few months ago, I made a dedicated Instagram account just to curate an algorithm around fitness and nutrition. I was annoyed that the algorithm on my personal account—full of baseball, hairstyle tutorials, fashion, bunny rabbits, and other forms of levity I desperately need during these trying times—was being overrun with high-protein recipes and exercise hacks. But the more I looked at the workout content my second account was suggesting, the more something stuck out to me: A lot of those videos really weren’t good.

Some were simply engagement-farming nonsense suggesting I do foolish, useless things in the gym. Others, though, seemed legitimately dangerous. I started looking at the content creators’ profiles and noticed a great deal of them weren’t even personal trainers or otherwise certified or educated in anything related to fitness. I consider myself someone who knows what to do in the gym for the most part, but some of these people were pretty convincing when claiming to know what they were talking about. Here’s what to look out for and avoid when you’re choosing at-home, guided workouts to follow.

Avoid anyone who isn’t clear about certification

I have absolutely no doubt there are plenty of people out there who have done their own research, toiled in the gym, and emerged from their personal wellness journeys super well-versed in all things health and fitness, all without ever taking a certification or education course in anything relevant. That does not mean you should listen to them. Even if 99 out of 100 are brilliant, right on the money with their advice, and knowledgeable about sports science, you always run the risk of inadvertently following the guidance of the one out of 100 who is not.

Certified personal trainers (CPTs), performance enhancement specialists (PESs), group fitness instructors (GFIs)—you’ll be looking for a lot of acronyms, basically, and they should be right there in the trainer’s bio on whatever platform you’re using. If someone is certified in something like yoga, pilates, or cycling, their profile should also say that outright. If you don’t see anything like that, keep it moving. Certifications are expensive and time-consuming; certified trainers advertise what they’ve earned upfront.

I’ve recommended a lot of my own favorite yoga, cycling, and pilates instructors whose classes are available on YouTube. The ones I selected all come from certified pros. As someone with a certification in teaching indoor cycling, I can’t tell you how different my classes are from what I imagined they’d be like before I took my training courses. Learning about the science behind different exercises and having to prove you retained it all with a certification test is crucial to providing safe, effective instruction.

One thing I do when I see a tip on Instagram but I’m not so sure about it is I just ask a trainer at the gym. If you have a gym membership, there are probably trainers available and most of them are happy to answer a quick question—especially if it can prevent someone from getting hurt. Bring them a coffee or something the next time you go in, as a thank-you (sugar and milk on the side, of course, so as not to offend their nutritional sensibilities).

Avoid poor production quality

I have so much respect for gritty content creators who want to share their expertise and, in most cases, I don’t care if a makeup tutorial or day-in-the-life vlog is shot on a noticeably bad camera or the audio is a little shaky. I commend them for having the wherewithal and the vision to try at all. That’s not the case with workout instruction, though, because poor quality can lead to unsafe practices.

One of the reasons I love the Peloton app so much is that the production quality is really high—which it should be, since I and millions of others are paying $44 a month for the service. It’s worth it, though, to be able to see and hear the instructors clearly so I never miss a form or safety cue.

That said, some of my favorite YouTube instructors have some shaky production and I won’t lie. I love Kristina Girod as a cycling coach, for instance, but I’ll admit her audio can be a little garbled from time to time. Avoiding poor production or unclear audio and video is important, but especially so when you’re a beginner and you need all the help you can get. As you progress, you might be able to get by without knowing precisely what resistance a cycling instructor is calling for or whatever, but you don’t want to make a habit of going rogue.

Avoid anything without variety or modifications

A one-size-fits-all approach isn’t very applicable when you’re setting out to work out on your own. A yoga class that’s easy for you might be hard for me and while it’s good that either of us is even trying it, it’s not that beneficial for you to be doing something that doesn’t challenge you at all, nor for me to be doing something I can’t do correctly. For something to be engaging enough for you to stick with it and also still be safe enough, it needs to be a little more tailored to your skill level.

In my pilates roundup, for instance, I only included instructors who either break down their videos by skill level or provide modifications throughout the workout by verbally explaining options people at different levels can try out. If you see content labeled “all levels” but the teacher offers no modifications, you’re better off spending more time looking for something that aligns with what you’re able to do, specifically.

The worst offenders for this one are found on short-form video apps like Instagram and TikTok. You see a lot of people demonstrate an exercise with no clear instruction, no modifications, and no context like how long they’ve been training that way or what weight they’re using. Replicate their vague advice (which is tailored just to their body and skills) at your own peril.

Avoid using only free instruction

I won’t insult your intelligence by advising you to avoid anything that advertises you can drop a certain amount of weight or gain a specific skill in a certain amount of time; you know that’s almost always nonsense. How, I ask you, can a three-week YouTube course help you lose precisely 21 pounds when it doesn’t know your starting weight or anything else about you? Be serious.

But that’s the problem with free content. Creators are incentivized to post sensational videos so they can monetize a higher amount of clicks. You might not want to hear this, but sometimes, you should avoid the free option. I do my best to find the top-quality free instruction that’s out there, but frankly, I love my Peloton and Les Mills+ subscriptions more than I love free YouTube videos. With those, I know I’m getting guidance from educated and credentialed trainers, the production quality is going to be high, and I’m not going to be served up any clicky gimmicks. This doesn’t have to break the bank. Peloton offers a version of its app to people who don’t have the company’s proprietary equipment for just $24 a month instead of the $44 it costs if you do have a Bike, Tread, or Row. A basic Les Mills+ membership is $15 per month.

A lot of the fitness instructors on YouTube, Instagram, and other platforms do offer paid-only content, as well. I recommend consuming their free content to feel out if they’re a fit for you, then considering subscribing.

OpenAI’s latest model, GPT-5, is officially here. The company announced the model in a “longer than usual” presentation today. The event, which spanned nearly an hour and a half, ran through some of the specific changes that come with GPT-5, as well as some practical use cases, though some of this information was spoiled in a Thursday morning leak.

Introducing GPT-5

First things first: OpenAI is clearly hyped about GPT-5, as any company with a new product would be. But some of the language here is interesting. OpenAI CEO Sam Altman likened the different AI models, or GPTs, to people of different ages and skill sets: Altman said GPT-3 is like talking to a high school student; GPT-4 is like talking to a college student; and GPT-5 is like talking to an “expert,” a keyword thrown around quite a bit during this presentation.

As with any new model, OpenAI says GPT-5 out performs previous GPTs and models on the market across a number of benchmarks, including SWE-bench, Aider Polyglot, MMMU, and AIME 2025. What piqued my interest in particular are OpenAI’s claims that GPT-5 has a far lower hallucination rate than previous models. AI models have a bad habit of hallucinating, or, in plain terms, making things up. OpenAI says GPT-5 makes up information less than other models, which will be interesting to put to the test.

Speaking of which, the model will be available to test starting today, Aug. 7. What’s more, OpenAI is making the full GPT-5 model available to all users—including those who use ChatGPT for free. Free users do have the lowest use limit when it comes to GPT-5, and when they hit it, they’ll be bumped to GPT-5-mini, a lightweight version of the model. Plus users, those who pay $20 per month for ChatGPT, have higher limits than free users, but will still be kicked to the mini model when they hit their limits, while Pro users ($200 per month) have no limits, as well as access to GPT-5 Pro. OpenAI didn’t share what those limits are in the presentation, though they may reveal that on their website in a separate post.

One key feature of GPT-5 is its ability to switch between a thinking and non-thinking model depending on the prompt. If you ask ChatGPT something simple, GPT-5 will respond with a quick, but detailed, answer. Ask it something more complex—say, a request to create something—and the thinking model kicks in. You’ll then be able to see the model break down the request step by step, and watch as it reasons through its response. If you pay for ChatGPT, you can access the thinking model from the model picker, but all users can request that GPT-5 think during the prompt.

OpenAI demonstrators casually announced that GPT-5 deprecates older GPT models, which means you likely won’t be able to use older models like GPT-4o or GPT-4.1 going forward. As such, demonstrators asked both GPT-4o and GPT-5 to write a eulogy for older models, showing how GPT-5’s writing is less cookie cutter than 4o and earlier. I disagree, though. While GPT-5 might have a bit more personality and flourish than GPT-4o, its writing had plenty of the telltale signs of AI-generated text, including flowery language and awkward similes.

Models

As with previous GPTs, GPT-5 comes in three different models. These include:

• GPT-5: This model is for logic and multi-step tasks.

• GPT-5-mini: This model is not as powerful as GPT-5, and, as such, is designed for “cost-sensitive” applications.

• GPT-5-nano: This model is designed to be fast for “low latency” applications.

This isn’t a huge surprise, since the news was leaked in an accidental GitHub post early Thursday morning.

Coding

My take is that OpenAI is particularly happy with GPT-5’s coding abilities. Much of the presentation was spent on demoing how the model can help write code, craft programs from scratch, and hunt for bugs. In the first example, a demonstrator asked the bot to create a website to help their partner learn French. They wanted GPT-5 to create three elements: a flashcard section; a quiz; and a game; specifically a recreation of Snake, but with a mouse hunting cheese, and anytime the mouse reached the cheese, the game would speak a word in French out loud.

Sure enough, GPT-5 generated the site. In fact, the demonstrator had it generate three different versions of the site, as a way of showing the variety of GPT-5 outputs. The sites were fine, if not a big plain—though don’t get me wrong, it’s incredible an LLM can do this at all, in just minutes. I’m not sure we’re anywhere close to being able to describe a high-quality website and expect a chatbot to spit it out in a minute, but this doesn’t make me want to invest much time in learning how to code.

In another example, the demonstrator asked ChatGPT to build a balloon popping game set around a castle on a mountain. Instead of waiting for the bot to generate the game, they pulled up a previously generated game, which means you should take this with a grain of salt. But sure enough, OpenAI demoed a crude 3D game which let you shoot balloons around a castle on a hill. You could even “chat” with the soldiers and characters around the castle, though in practice it was more of a simple chatbot function with different personalities.

Deception and safety

OpenAI touched on GPT-5’s deceptions, or, in other words, its ability to lie. This is different from hallucinations, as deceptions seem more deliberate—rather than simply making up information, the bot actively attempts to deceive you, perhaps in response to news that you’re shutting it down. Scary stuff.

OpenAI claims that GPT-5 is less deceptive than both o3 and o4-mini. Plus, if you ask it a question that would normally be refused by a previous model due to safety concerns, it’ll try to answer the question in a impartial or objective way—in an attempt to service those who are asking legitimate questions. If it must refuse, it’ll give you a detailed answer as to why.

Health

Sam Altman was particularly keen on hyping up GPT-5’s health capabilities. Altman called GPT-5 the “best model ever for health,” and that it scores higher than any previous model on OpenAI’s HealthBench benchmark.

Altman then brought out a couple to discuss their use of ChatGPT in navigating a difficult and complex cancer diagnosis. While most of their experience was using previous ChatGPT models, the couple did note that GPT-5 was particularly helpful in going deeper on their questions—giving them additional information, including questions they should ask the doctor or what to expect next.

Other new features

Demonstrators also touched on Voice Mode, but not for long. The major demo was of Study and Learn Mode, which supposedly helps you learn subjects using Voice Mode. They demoed this by asking the bot to help them learn Korean, specifically to slow down its speech so they could hear each individual word clearly. The bot did, though it didn’t slow down quite enough for me, someone who does not speak Korean, to understand. Once GPT-5 hits my ChatGPT account, I’ll be curious to see if you can ask the bot to speak even slower.

Additional, smaller features announced include the ability to customize the color of your ChatGPT conversations, as well as Gmail and Google Calendar integration. The latter gives me privacy concerns, but I imagine any Google Workspace users who also rely on ChatGPT will enjoy this collaboration.

Disclosure: Lifehacker’s parent company, Ziff Davis, filed a lawsuit against OpenAI in April, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.

We may earn a commission from links on this page.

Even though it costs me a borderline comedic amount of money to take Pilates classes at a reformer studio here in Manhattan, I’m quite willing to pay—I love Pilates that much. And even though I’m also fortunate enough to work at a gym where we have free mat Pilates classes, I still want more, which is how I find myself perusing YouTube and my beloved Peloton app for guided classes I can follow along with at home. I’m a sicko, but it’s fine.

Luckily, for my fellow sickos (and the more well-adjusted Pilates fans out there), there are plenty of great, free Pilates instructors available on YouTube. You can use these classes to familiarize yourself with Pilates if you’re new, save some money if studios around you are getting too expensive, or stick to your routine when you’re away from traditional classes.

Move With Nicole

This channel has 5 million subscribers, which is a pretty solid endorsement on its own. I like Move With Nicole because the instructor produces such a solid variety of content. From the title of each video, you’ll immediately know if it’s easy or more advanced, how long the routine will take, and whether you’ll need equipment. (Personally, I don’t have a lot of dedicated Pilates equipment in my home, since I can’t justify investing in it when I have access to so many in-person classes that provide it, so I’m always on the lookout for YouTube workouts that don’t require anything too fancy.)

Nicole McPherson is certified to teach Pilates and yoga, both of which are available on the channel, which rounds out the offerings nicely. On days you might be more interested in a calmer stretch, she gives you that option through her yoga, all without you having to go hunt elsewhere. She speaks in soothing, calm tones while she gives clear instruction about what to do (which is really valuable, especially for beginners) and as a bonus, you get to see some gorgeous shots of her environment in Thailand.

Flow With Mira

On Flow With Mira, you can find videos broken down by skill level, but also ones dedicated to routines with specific equipment, so if you do happen to have a magic circle or Pilates ball, you can really put it into action here. Videos are also broken down by muscle group, which makes it simple to build a weekly program around the channel.

Mira Bassan leads the classes for her 244,000 subscribers with clear, direct instructions and little fanfare. She tends to jump right into routines after explaining which muscle groups will be targeted and her production is solid enough that you can see clearly exactly what she’s doing as she describes it.

Trifecta Pilates

Trifecta Pilates, which has 209,000 subscribers, is a good place to turn when you don’t want to get decision paralysis about which sort of class to follow. Beth Sandlin offers up all-level routines, meaning just about anyone can tune into one of her videos and get something out of it. Rather than creating workouts for various skill levels, she provides modification options throughout her routines so everyone can follow the same clip.

In addition to classes of various lengths and using different types of equipment, she also provides educational content for other instructors or just people curious about Pilates, hosting discussions with other experts or giving overviews of Pilates’ history.

Blogilates

No list of at-home Pilates channels is complete without a nod to Blogilates, maybe the most recognized name in the game, thanks to the instructor’s lines of activewear and equipment through Target and her brand Popflex. Cassey Ho has nearly 11 million subscribers and has been in the game a long time. (I was watching her videos 14 years ago.) While she’s moved on to vlogging about her experiences as an entrepreneur, plus nutritional content and other standard YouTuber fare, even the old Pilates videos are great. (You can track how much production improved over the years, too, but they’ve always been good workout videos, even when the quality wasn’t that high.)

Ho has structured the channel meticulously. There is a new workout playlist curated every day, so you can just tap on the one designed for whatever day it is and jump right into her program. Her instructions are clear and she’s super upbeat, creating some contrast with the other, more zen instructors on the platform.

Jessica Valant

With over 1 million subscribers of her own, Jessica Valant is a high-quality instructor who really knows her stuff. Not only is she certified to teach Pilates, but she’s a physical therapist. She is especially knowledgeable about women’s bodies and structures her workouts around her expertise.

Her channel is easy to navigate and videos are divided into categories based on their length, type, muscle focus, and more. Valant is dedicated to making her routines accessible, so her approach is educational and encouraging all at once. She’s clear when she gives instructions, safety cues, and modifications, so you can follow along easily.

We may earn a commission from links on this page. Deal pricing and availability subject to change after time of publication.

If you’ve been looking for open-ear headphones that let you hear the outside world without totally sacrificing sound quality, these JBL SoundGear Sense are worth checking out. They’re currently on sale for just $74.99 on StackSocial, refurbished and listed in “Grade A” condition, so you’re getting near-mint hardware for a pretty steep discount. These aren’t your typical noise-isolating buds: They sit outside your ears, held in place with adjustable hooks, and come with an optional neckband for more stability when moving around.

The open design means your surroundings stay audible, which makes them ideal for running, biking, or any situation where awareness matters. That also means you’re not getting chest-thumping bass—lows are present, but not deep, and they tend to drop off when you crank up the volume, according to this PCMag review. Still, the sound is clean and punchy thanks to 16.2mm drivers, and the four built-in mics deliver surprisingly clear calls, even in windier conditions. Touch controls work fine for the most part, though they’re reportedly a little too easy to trigger by accident. You can tweak their sensitivity and EQ using the My JBL Headphones app, which adds some welcome customization.

Battery life is solid: You’ll get around six hours per charge, with another 18 from the case. A 15-minute quick charge adds four more hours, which is great if you’re heading out the door in a rush. They’re also IP54 rated, meaning they’ll survive sweat and the occasional splash. Just note: there’s no active noise cancellation, but that’s by design. These are meant for transparent listening. If you’re okay trading isolation for awareness, and you like the idea of earbuds that don’t block your ears, the SoundGear Sense makes a solid, affordable pick, especially at this price.

The average American gets more than half of their calories from ultra-processed foods, according to a new federal report. That’s not surprising—ultra-processed foods are everywhere, and includes many of our cheapest grocery options. That’s why it’s concerning that ultra-processed foods have been linked to a variety of health problems, including heart disease, diabetes, mental health disorders, and everyone’s favorite, all-cause mortality.

But what counts as ultra-processed food? Let’s dig in, and maybe question a few assumptions along the way.

The new study, from the Centers for Disease Control, found that the most common source of ultra-processed food for adults was “sandwiches,” followed by sweet bakery products, sweetened beverages, savory snacks, and “bread, rolls, and tortillas.” For kids, the findings were similar, with pizza and savory snacks outranking sweetened beverages. For kids, bread didn’t make the top five. The study also found that consumption of ultra-processed food had actually dropped a bit from 2017 to 2023, the last year for which data was available. You can see the summary of results here.

Before we get too into the weeds here, there’s a big caveat worth remembering: When studies look at the health of people who do and don’t eat ultra-processed foods, they’re not necessarily studying the fact that the food is ultra-processed. We can say pretty definitively that a diet high in candy bars is bad for you, but is that because the candy bars are ultra-processed, or because they’re full of sugar? That’s not a question that the current research can really untangle, but it’s important one to consider. Are ultra-processed foods always bad, or are they just a category that includes a lot of foods we should eat less of?

How are ultra-processed foods defined?

This terminology comes from a classification scheme called NOVA that splits foods into four groups:

• Unprocessed or “minimally processed” foods (group 1) include fruits, vegetables, and meats. Perhaps you’ve pulled a carrot out of the ground and washed it, or killed a cow and sliced off a steak. Foods in this category can be processed in ways that don’t add extra ingredients. They can be cooked, ground, dried, or frozen.

• Processed culinary ingredients (group 2) include sugar, salt, and oils. If you combine ingredients in this group, for example to make salted butter, they stay in this group.

• Processed foods (group 3) are what you get when you combine groups 1 and 2. Bread, wine, and canned veggies are included. Additives are allowed if they “preserve [a food’s] original properties” like ascorbic acid added to canned fruit to keep it from browning.

• Ultra-processed foods (group 4) don’t have a strict definition, but NOVA hints at some properties. They “typically” have five or more ingredients. They may be aggressively marketed and highly profitable. A food is automatically in group 4 if it includes “substances not commonly used in culinary preparations, and additives whose purpose is to imitate sensory qualities of group 1 foods or of culinary preparations of these foods, or to disguise undesirable sensory qualities of the final product.”

That last group feels a little disingenuous. I’ve definitely seen things in my kitchen that are supposedly only used to make “ultra-processed” foods: food coloring, flavor extracts, artificial sweeteners, anti-caking agents (cornstarch, anyone?) and tools for extrusion and molding, to name a few.

Are ultra-processed foods always bad?

So we’ve learned that packaged snack cakes are ultra-processed, and so is a factory-baked loaf of bread that has 20 ingredients. Orange juice whose flavor has been manipulated would count, too. Coke and Diet Coke are both solidly in this category. It seems logical that we should eat less of these things.

But you could argue that the real problem with these foods is that they’re often sugary and high calorie, and many of the less-healthy members of the category are what stock the vending machines and convenience stores that beckon to us when we’re hungry and haven’t packed a lunch. The problem with these foods is that a diet full of them is unbalanced, due to the nutrition they do or don’t contain. The processing itself isn’t the problem.

So when we talk about ultra-processed foods, we have to remember that it’s a vague category that only loosely communicates the nutrition of its foods. Just like BMI combines muscley athletes with obese people because it makes for convenient math, NOVA categories combine things of drastically different nutritional quality.

Why the level of processing isn’t always the most important thing

Illustrating the point above, the USDA published their own study showing how you can create a healthy diet out of ultra-processed foods. A homemade breakfast burrito, for example, might contain canned beans, liquid egg whites, shredded cheese, and a store-bought tortilla. Those ingredients might be ultra-processed, but they’re nutritionally nothing like grabbing a Cinnabon on your way to work.

A pet peeve of mine is that the NOVA classification sometimes draws distinctions between things that aren’t really nutritionally different. Wine is in group 3 next to cheese and fresh bread, but cocktails are in group 4 with the Twinkies. Hard liquor has been distilled, you see, so it’s ultra-processed.

Canned vegetables are in group 3 (processed) while their fresh counterparts are in group 1. But canned veggies aren’t any less nutritious. Meanwhile, dried fruit is in group 1 (so wholesome!) even though it can be more sugary than cakes or cookies.

There’s a lot of overlap between unhealthy(?) foods and ultra-processed foods, so I understand why scientists are studying ultra-processed foods as a group. But demonizing UPF, as they’re sometimes called, often ends up putting the cheapest, most widely available food in the most shameful category. Is that fair, or does it just make you feel better when you’re eating fresh green beans and scoffing at people who buy canned?

The NOVA scale isn’t totally useless: It helps researchers keep an eye on how much of our food is coming from large-scale manufacturers. But it’s not a great way to evaluate what’s in our grocery bags, or on our plates.