• Meta denies plans for new performance-based layoffs amid online speculation.
• Meta previously considered annual job cuts based on performance to manage low performers.
• Meta recently cut 10% of its Reality Labs division, affecting over 1,000 employees.

Meta says it will not have a fresh round of performance-based layoffs, even as a smattering of online chatter has raised questions about whether the social media giant will quietly restart its performance-driven purge.

“These are individual cases not related to any company wide initiatives,” a Meta spokesperson told Business Insider when asked about a recent restructuring. “For example we are not doing any 5% low performers like we did last year.”

That’s a notable shift in tone from early 2025, when Business Insider reported that an internal FAQ circulating at Meta suggested performance-based job cuts could become an annual practice, with the company saying it “may use future performance cycles” to move out its lowest performers. Early last year, Meta cut 5% of its workforce, saying it was focusing on its lowest performers.

The clarification also comes as Meta continues to reshape other parts of the business. Last month, the company cut about 10% of its Reality Labs division, a move that affected over 1,000 employees.

Have a tip? Contact Pranav Dixit via email at pranavdixit@protonmail.com or Signal at 1-408-905-9124. Use a personal email address and a nonwork device; here’s our guide to sharing information securely.

• The US Army is working on an artificial intelligence system to help process more data than humans could.
• The AI is fast, but it also remembers the context and connections humans might forget or miss.
• Soldiers are continuing to develop the AI but have seen success with it so far.

Army leaders say the modern battlefield is so saturated with sensors and networked weapons generating more data than soldiers can realistically process on their own that artificial intelligence is needed to meaningfully sort it all.

For years, the Army’s focus was on fielding more sensors for battlefield information and awareness, but now the service is also having to think about information overload and managing the massive amounts of data coming in.

During a recent US Army and NATO exercise in Europe, troops used a homegrown AI system to consume and sort data. The value wasn’t strictly that the AI could do it faster but rather that it could remember context and patterns that humans couldn’t.

The case from the Dynamic Front exercise is another example of how the US military is increasingly implementing AI and automation into everything from enemy attack simulations to paperwork.

“The modern battlefield, what we’re already seeing across the globe, it is swimming in sensors, and we are drowning in data,” Col. Jeff Pickler, the Army 2nd Multi-Domain Task Force commander, said at a media roundtable on Dynamic Front.

There aren’t enough people to decipher all the available information, he said. “They will never be able to fully process all of that.”

The software aimed at addressing that problem remains in beta testing. In the next iteration of Dynamic Front — which will merge with another exercise, Arcane Front, to pair technology experimentation with theater-level combat rehearsals — Army leaders say they intend to test the AI at a larger scale.

“If we’re looking at a target set in the European theater where we think we’re going to need to process upwards of 1,500 targets a day, that’s beyond the human scope,” Pickler said. “The answer to the equation there is in AI and automations.”

During a potential large-scale conflict in Europe, AI could assist in locating and assessing those targets.

The system can do this quickly, but the speed isn’t the main benefit. AI can remember patterns that humans might forget or not even notice. Pickler gave an example of AI realizing that unrelated shipping reports, a local power outage, and a fertilizer delivery together might suggest missile fueling activity.

“So the difference isn’t seconds versus minutes — it’s minutes instead of months. Not because the machine scans quickly, but because it keeps context across sources that humans can’t hold in memory,” Pickler said after the roundtable.

“It doesn’t replace analysts by reading faster,” he said, “it replaces the weeks analysts spend reconnecting information spread across thousands of reports.”

In a conflict scenario, that could mean analysts reach a clearer picture of the battlefield faster. Correlations between data gathered from different sensors could surface more quickly. If an adversary were fueling, arming, or moving weapons in ways that were not immediately obvious, AI could help flag those links.

Humans, though, would still decide how to respond.

Soldiers have seen success with iterating on the current AI model, the Army said. It’s been retooled during testing, and humans remain in the loop, reviewing outputs at multiple stages.

The goal is to continue increasing the overlap the model would have with human-produced information. In a targeting example, a milestone would be if AI achieved 90 to 95% agreement with humans on 100 target sets.

The Army’s push for AI and automation is also driving the development of its Next Generation Command and Control software, a priority initiative.

The technology being developed by vendor teams including Anduril, Palantir, and Lockheed Martin uses AI and machine learning to provide commanders and soldiers with real-time data on ammunition levels, maintenance needs, intelligence feeds, targeting, and simulated enemy attacks.

But AI is also changing other aspects of how the Army works. Autonomous features in drones, weapons, and targeting might be at the forefront, but behind the scenes, personnel are using new tools, redesigned workflows, and data integration for recruiting, maintenance, and inventors. These are manual tasks that the service believes can be improved with AI.

• Gary Marcus said Matt Shumer’s viral essay about the future of AI missed the mark.
• The AI researcher said Shumer didn’t grapple with the current limitations of AI models.
• “I think that one should work from the facts rather than just trying to cause an alarm,” Marcus told Business Insider.

If that viral essay about AI had been printed on paper, there’s a good chance AI researcher Gary Marcus would’ve ripped it up in disgust.

Marcus acknowledges something is happening in AI — just nowhere near the scale described in the recently viral essay, which predicted a looming disruption “worse than COVID.”

Marcus, who on X criticized the essay written by entrepreneur and investor Matt Shumer as having “not a shred of actual data,” dismissed its contents as alarmist in an interview with Business Insider.

“Hyped-up views have gotten us into a bad place, possibly one that’s going to lead to a serious economic recession or something like that,” Marcus told Business Insider. “And I guess I think that one should work from the facts rather than just trying to cause an alarm.”

In his essay titled “Something Big is Happening,” Shumer, whose past startup sells a subscription-based AI-assisted writing tool, warned that AI would upend not just software engineering, but most jobs done “on a screen.” Shumer also has a small VC fund.

Marcus said that while AI will replace some labor and affect jobs, the process will be much slower than what Schumer and others are describing.

AI can do some things well and help speed up work, but it’s just not near the point of replacing humans, Marcus said.

“AI can do a small subset of the tasks, and that sometimes speeds up human beings and things like that, but it rarely does all of what a human being can do in any particular domain,” he told Business Insider. “This will change over time, just to be clear. It is likely that AI will replace most human labor over the next century, but it’s not likely that it will over the next year or two.”

Companies that move too quickly to replace jobs with AI are likely to find themselves in a similar position as Klarna, Marcus said. In 2024, Klarna touted an AI assistant that could do the equivalent work of 700 people. By May 2025, CEO Sebastian Siemiatkowski, long a proponent of AI, said the Swedish fintech was leaning back into recruiting actual people.

“Six months or a year later, they come back with their tails between their legs because it turns out that the AI systems don’t do things as well as the human,” Marcus said. “So, I’m not saying that there’s nothing going on. I’m not saying that there’s no value in these AI systems, but they’re premature.”

Marcus said that the more likely outcome in the short-term is not that AI will replace junior employees but rather that executives think it’s capable of doing so — and make what could ultimately prove to be a costly gamble.

“The biggest thing I think junior people have to worry about right now is a misapprehension by the C-suite that these techniques work better than they actually do,” Marcus said.

As of Friday morning, Shumer’s post has been viewed more than 80 million times on X alone. In a Substack post expanding on his criticisms, Marcus called Schumer’s post “weaponized hype.”

“The general impression that he conveys is basically that the sky is falling now, and at most, I think what’s really happening is the junior people are under some threat, and I think that threat is actually exaggerated,” Marcus told Business Insider.

Shumer previously told Business Insider that he wrote his essay in part to reach people like his dad, who may be skeptical or avoid AI entirely. He felt compelled to warn them about what might be on the horizon, even “if there’s just a 20% chance of it happening.”

Marcus’s biggest critique of Schumer’s post is that it doesn’t take into account current data and research showing that AI still has a long way to go, and that it didn’t present the full context behind a famous Model Evaluation & Threat Research graph assessing AI progress.

He said that other studies, including a June 2025 paper published by Apple’s Machine Learning Research Group, found limitations in what current models can do.

Marcus also said that many leading AI CEOs who have made bold predictions about the future of work have failed to deliver on past ones. He points to xAI CEO Elon Musk’s frequent rosy outlook on the number of robotaxis Tesla will put on the road (Musk once said one million by 2020) and to Nobel laureate Geoffrey Hinton’s 2016 statement that the world should stop training radiologists. (Last May, Hinton told The New York Times that his prediction was poorly worded and that while he was wrong on the timeline, the general direction for AI’s capabilities in radiology was correct.)

“What they have all learned to do is to sell the rosiest picture possible, and the media rarely calls them out,” Marcus told Business Insider.

On Thursday, Microsoft AI CEO Mustafa Suleyman made waves by predicting that most, if not all, white-collar tasks could be automated within the next year and a half.

One of the industries Suleyman mentioned is accounting. Marcus isn’t convinced.

“Think about accounting in particular,” he told Business Insider. “Even one mistake can cost a client hundreds of thousands of dollars or get them sent to jail or whatever. Accounting is a business that is built on accuracy. If you’re not accurate, you don’t have a business.”

• New York Fashion Week runs from February 11 to February 16 this year.
• Celebrities including Anne Hathaway, Pamela Anderson, and Elle Fanning have attended so far.
• The best looks have sparkled and stood out, while some of the worst were mismatched and understated.

The best of fall and winter fashion is officially on display in the Big Apple.

The first New York Fashion Week of 2026 kicked off on February 11 and will continue through February 16.

Designers like Ralph Lauren and Coach have showcased their new styles, while celebrities, including Anne Hathaway, have made fashionable appearances at runway shows.

Here’s a look at the best and worst A-list looks we’ve seen.

Anne Hathaway was stunning in a black Ralph Lauren gown.

She attended the designer’s fall runway show wearing a high-neck halter gown made from gauze-like fabric.

It wrapped around her body, creating a subtle see-through effect, and its skirt extended into a short train.

She wore it with a fuzzy jacket that she carried around her waist.

Pamela Anderson’s black-and-white outfit missed the mark.

She made an appearance at the Tory Burch fall/winter runway show wearing a preppy look from the brand. It comprised a white pleated skirt with a black, button-up blouse tucked in.

Unfortunately, the two pieces sat loosely on Anderson and appeared too formal.

Her leather belt, sheer tights, pointed heels, and black purse added some fun to the look, but the base layers were ultimately too simple for them to work.

Lili Reinhart was chic in a statement skirt.

While entering the Ralph Lauren show, Reinhart was photographed wearing a brown tweed blazer as a top with a thick brown belt holding it closed at the waist.

She also wore a long black skirt covered in leather fringe, which added texture and contrast to the ensemble.

High black boots and a small brown purse completed the fun, fashion-forward look.

Ana Amelia Batlle Cabral had a sharp fashion moment, but Marcello Hernández’s look needed one tweak.

The couple attended the Tory Burch show together in contrasting outfits.

Cabral, an architect, wore a blue pleated skirt atop a darker, unbuttoned blazer, and brown Tory Burch heels with silver embellishments. The look was stylish and sharp.

Then there was Hernández, the “Saturday Night Live” comedian, who opted for khaki pants, suede sneakers, and a two-toned sweater.

Though the latter two pieces looked great on him, his trousers could have benefited from a bit of steaming and tailoring.

Amanda Seyfried’s mixed metals worked for her.

Also at the Tory Burch show, Seyfried wore a golden, long-sleeved blouse atop a blue midi skirt with silver stripes. She also wore white heels with silver adornments and carried a blue purse.

The mix of two metallic shades with blue tones created a unique, fun color combination that complemented her hair and skin tone.

Suni Lee’s multicolored look overwhelmed her.

Outside the Tory Burch runway show, the Olympic gymnast was photographed wearing an oversize blue jacket with a form-fitting wrap skirt in a vibrant red shade.

The two pieces were drastically different in style, and they overwhelmed her petite frame. They also didn’t match the lime-green top she wore or her black accessories.

Caleb McLaughlin was cool and casual at the Coach show.

The “Stranger Things” actor attended the show in blue jeans, a button-up top, and a leather jacket that matched his sneakers and Coach purse.

The latter piece was especially fun. It featured a prominent pocket reminiscent of a coin purse and two keychains that looked like miniature books.

McLaughlin has been one of the best-dressed men at New York Fashion Week so far.

Elle Fanning’s outfit had potential, but it needed a change.

Fanning entered the Coach show in a white gown with a mesh, star-print overlay. It was fun, feminine, and the perfect piece to pair with her cropped leather jacket and brown purse.

That said, she also wore thick white boots with the outfit. A simple pair of heels or flats would have been more in line with the look.

Kelsey Merritt stood out at the Carolina Herrera show.

The model sat in the stands wearing skinny black pants with a billowing blouse. The semi-sheer white garment had a high neckline, balloon sleeves, and extra fabric at the waist that extended into a train.

The top piece was dramatic and memorable, while her pants were perfectly tailored to highlight her dramatically pointed heels.

Ben Platt needed more color variety in his New York Fashion Week look.

For the Michael Kors runway show, the actor wore an all-white ensemble that included trousers, a loose-fitting blouse, sneakers, and a trench coat.

The outfit would have worked for him if it featured more than one color. Because the entire ensemble was stark white, each piece blended into the others, creating the illusion that he was wearing a baggy jumpsuit.

• The internet is again abuzz with a new Chinese AI model.
• Seedance 2.0, ByteDance’s new video-generation tool, can create hyperrealistic videos.
• Users are making videos of Hollywood actors, raising copyright concerns.

In a new viral AI video, Brad Pitt and Tom Cruise pummel each other on a rooftop in a cinematic action sequence.

It’s not a trailer for a new blockbuster, and it’s not actually Pitt and Cruise, though it looks a lot like them. The video is so realistic, in fact, that the clearest sign it’s made with AI is the dialogue.

“You killed Jeffrey Epstein, you animal! He was a good man!” Pitt says as he punches Cruise.

“He knew too much…” Cruise replies.

You can see why Hollywood’s most prominent trade organization is not happy about it.

The scene was created using Seedance 2.0, a new AI video-generation model released Thursday by ByteDance, the Chinese parent company of TikTok.

The tool and the hyper-realistic videos of Hollywood actors and characters that users are creating with it have gone viral in China and are now catching the attention of Americans.

“It’s happening fast,” Elon Musk said in response to a video generated using Seedance and posted to X, a reference to the speed at which artificial intelligence is advancing.

“We’re cooked,” another X user said.

The reaction from Americans is reminiscent of the buzz around DeepSeek, a Chinese company that unveiled an AI reasoning model in January last year that rivaled OpenAI’s ChatGPT and other top models, stunning the biggest names in Silicon Valley and ratcheting up the competition between the United States and China in the race to dominate AI innovation.

Seedance 2.0 is another shot across the bow of American AI companies. Its multimodal capabilities span text, image, audio, and video inputs and give creators control over metrics such as lighting, shadows, and camera movement.

In another viral scene, a deepfake version of Walter White — of Breaking Bad fame — points directly at the viewer and says, “You think you’re in control.” It’s a line that feels less like dialogue and more like a taunt.

The uncanny representations of Hollywood actors, as well as characters from the Avengers and other major movie franchises, immediately raised copyright concerns.

The Motion Picture Association, the trade group representing major Hollywood studios and streaming services, released a statement on Thursday accusing ByteDance of infringement on a “massive scale” in a “single day.”

“By launching a service that operates without meaningful safeguards against infringement, ByteDance is disregarding well-established copyright law that protects the rights of creators and underpins millions of American jobs. ByteDance should immediately cease its infringing activity,” Charlie Rivkin, the chairman and CEO of the MPA, said in a statement.

OpenAI’s AI video generation tool, Sora, which can also create AI versions of actors and characters, has also raised copyright issues.

Like with so many of AI’s latest tools, however, it can be hard to put the genie back in the bottle.

• JPMorgan said it’s reorganizing its commercial and investment bank to “maximize” AI in an internal memo.
• The bank appointed Guy Halamish as the chief operating officer of the CIB as part of the effort.
• In his role, Halamish will oversee a group of new data and analytics officers driving the AI push.

JPMorgan is consolidating power to move faster on AI.

The bank is reshuffling its commercial and investment bank to “maximize the impact of AI,” according to an internal memo seen by Business Insider that was sent this week.

The firm has named Guy Halamish as the chief operating officer of the CIB and tasked him with overseeing the ongoing effort to “harness the power of our data and fully leverage rapidly evolving AI capabilities,” the memo, signed by the CIB’s co-CEOs, Doug Petno and Troy Rohrbaugh, said. Halamish’s new role was first reported by Bloomberg.

Under the new structure, each major business in the division, including banking, markets, payments, and securities services, will have its own chief data and analytics officer reporting jointly to Halamish and business heads. The bank recently hired Zachery Anderson as the chief data and analytics officer of its payments division, after a nearly six-year stint at UK-based lender NatWest. In a LinkedIn post about the new job, Anderson said he wants to push the “edge of the possible with AI.”

The move is part of a new strategy to break silos across the unit and speed up adoption of AI.

The team of officers will work with the wider firm on a range of efforts, including “preparing our infrastructure for more advanced AI and the expanded use of AI agents” and “driving end-to-end transformation” in areas such as client onboarding.

The CIB is a huge profit driver for JPMorgan — in 2024, it generated $25 billion in net income out of a firmwide total of $58.5 billion, according to that year’s annual report.

JPMorgan, backed last year by an approximately $18 billion tech budget, is one of the financial industry’s leaders in AI, with its own proprietary genAI platform and additional tools in the pipeline. CEO Jamie Dimon defended the firm’s AI spending on a recent earnings call.

“We are going to stay out front, so help us God,” Dimon said about the spending.

Work at JPMorgan or have a tip? Contact this reporter via email at atecotzky@insider.com or Signal at alicetecotzky.05. Use a personal email address and a nonwork device; here’s our guide to sharing information securely.