Contact Me By Email

Friday, April 19, 2024

How TikTok Changed America - The New York Times

AN INCOMPLETE GUIDE

Love, Hate or Fear It,

​​TikTok Has Changed America

As lawmakers argue for TikTok to be sold, some of the app’s most popular memes, from skateboarding with a Fleetwood Mac soundtrack to the renegade dance, have been seen tens of millions of times.

Has there ever been an app more American seeming than TikTok, with its messy democratic creativity, exhibitionism, utter lack of limits and vast variety of hustlers?

And yet, of course, TikTok is not American, which is the whole reason that in March, the House of Representatives passed a bill with broad bipartisan support that would force the Chinese owners of the video-app juggernaut to either sell to a non-Chinese owner or face a ban. Lawmakers say it’s a national security threat, and that the Chinese government could lean on its owner, ByteDance, to obtain sensitive U.S. user data or influence content on the app to serve its interests.

Roughly 170 million Americans use TikTok. That’s half the population of the United States.

There’s a long road of legislation, deal making and legal challenges ahead before TikTok could be forced to change ownership or even be banned. The Senate would need to pass the legislation — which it may do as soon as next week, now that the House has bundled it into a foreign aid package that it is likely to vote on over the weekend. It would have to survive lawsuits from TikTok and creators. Buyers would have to clear regulatory approval. And after all that, Beijing could simply block a deal. But imagining what a United States without TikTok would look like throws into sharp relief just how much the app has worked its way into American culture.

TikTok, which officially landed in the United States in 2018, was the most downloaded app in the country, and the world, in 2020, 2021 and 2022. It wasn’t that the elements of it were so new — compelling videos from randos had long been a staple of American pop culture — but TikTok put the pieces together in a new way.

Unlike Instagram, Facebook or Snapchat, TikTok didn’t build itself around social connections. Its goal is pure, uncut entertainment. The algorithm ingested every data point it could from what users skipped, liked or shared — and spat it directly into the maddeningly habit-forming For You Page. Fans whispered reverently that it knew them better than they knew themselves.

Here are 19 ways of understanding how TikTok became part of American life. The music America listens to, the movies it sees, what conspiracies it believes, how it can make or break a product’s success, who it defines as a celebrity — all of it has been influenced by TikTok, for good and bad. Even if you’ve never opened the app, you’ve lived in a culture that exists downstream of what happens there.

ItBecameHollywood’sFavoriteMarketingMachine

Insular, slow-changing Hollywood responded to TikTok’s arrival in 2018 in typical fashion: complete dismissal. We’re way too busy making pictures to worry about some new short-form video app.

Then came denial. (This thing is just another fad.) Next, fear. (Teenagers and young adults are never going to the movies again!)

But there’s a plot twist: Hollywood has come to see TikTok as indispensable. 

“Anyone But You,” a Sony romantic comedy starring Sydney Sweeney and Glen Powell, arrived to a piddly $8 million in ticket sales over Christmas weekend. The movie turned into a full-fledged hit ($219 million) after TikTok users (at the urging of Sony) began making videos of themselves re-enacting the credit sequence. 

TikTok also served as a ticket-selling machine for “M3gan,” a Universal-Blumhouse horror movie about a sassy robot that has spawned a new franchise; “Wonka,” which debuted in December and collected $632 million; and the Barbenheimer box office phenomenon, otherwise known as “Barbie,” with $1.4 billion, and “Oppenheimer,” with about $1 billion. Rote glamour shots and insipid interviews — ye olde studio publicity tools — don’t work on TikTok; users want behind-the-scenes “realness.” Hence “Oppenheimer” stars goofing in a hotel hallway before a premiere, and pink-clad “Barbie” stars cavorting on the floor with puppies.

“Now that studios have figured out how to harness TikTok, the last thing they want is for it to go dark,” said Sue Fleishman, a former Universal and Warner Bros. executive who is now a consultant. “That would actually be a big problem.”

It’sGenZ’sWalterCronkite

Recently, V Spehar has posted TikTok videos telling viewers what they might have missed from President Biden’s State of the Union address, the first 15 actions that former President Donald J. Trump said he would take if he’s re-elected in November and Caitlin Clark’s WNBA starting salary.

Mx. Spehar posts to more than three million followers from the handle @UnderTheDeskNews and films many clips lying on the floor, a gimmick that began as an effort to differentiate from the authoritative tone of traditional television news anchors. The style of communication has resonated enough to make Mx. Spehar a regular at White House briefings with social media influencers.

News aggregation and analysis accounts like Mx. Spehar’s are shaping the discourse about current events in the United States, especially among young people. They’re a modern version of old-school bloggers — users respond to the personal tone, and the editorializing. (Some creators have even built followings simply by reading print news articles to their followers.)

Pew Research Center has found that about one-third of 18- to 29-year-olds say they get news regularly on the platform, far outpacing people in other age groups. 

In 2023, about 14% of American adults said they regularly got news on TikTok, compared with just 3% percent in 2020.

Other sites have similar draws. Roughly 16 percent of all American adults get their news from Instagram, and a similar amount from X. Far more people consume news on Facebook and YouTube.

The appeal of TikTok and other social sites has made mainstream outlets nervous, and has raised some concerns around accuracy and context as original reporting is funneled through other accounts. The Wall Street Journal has more than 340,000 followerson TikTok, while The New York Times has nearly 630,000 — numbers that pale in comparison with the followings of individual commentators like Mx. Spehar.

ItSuperchargesConspiracyTheories

Several dentists recently took to TikTok to debunk a conspiracy theory: that toothpaste tubes were printed with secret codes signaling their true ingredients to powerful people in the know.

Their efforts garnered far fewer views than the video that offered up the theory in early January. Not counting all the times the post was referenced in videos by other TikTok users, it has been seen more than seven million times in less than three months.

Tall tales are common on TikTok, where a flimsy patchwork of assumptions and coincidences — often concerning the schemes of a nefarious echelon of elites — is illustrated by dramatic images generated by artificial intelligence and spooky musical tracks. (Other such hits include false theories that President Joe Biden rigged the Super Bowl in favor of the Kansas City Chiefs or that Justin Bieber had signaled he was a victim of PizzaGate. False allegations of voter fraud also abound.)

Abbie Richards, a misinformation researcher who studies the TikTok ecosystem, said that such posts thrive because of the platform’s potent recommendation algorithms and its low barrier to entry.

TikTok allows users to earn money from their videos through tools such as its creator rewards program and livestream subscriptions. Conspiracy theories, which draw high engagement, are one of the most profitable categories, said Ms. Richards, a senior video producer at the liberal watchdog group Media Matters.

“It’s like candy for your brain — it tells a story that simplifies the world in a way that feels good to you,” she said.

A quarter of American adults who use the app create 98% of its videos.

The toothpaste theory was promoted by two young men known for conspiratorial content, including popular posts about satanic hit men and Britney Spears. They claimed that the colored dots on toothpaste tubes correspond to all-natural, medicinal or chemical ingredients.

The post was quickly reposted, copied and stitched into reaction videos. Some came from dentists, who explained that the dots were actually used during the toothpaste packaging process to help guide manufacturing equipment to properly cut and seal the tubes.

That conspiracy theory is not new — they rarely are on TikTok. Colgate, a major toothpaste manufacturer, addressed the color patch rumor last year and said that “as much as we love cracking secret codes, this one actually has nothing to crack because it’s entirely untrue.”

Even silly rumors, however, can spin out from TikTok into real-world harms. The baseless concerns that store-bought toothpaste tubes might hide toxic ingredients reignited recommendations to opt instead for unproven and potentially damaging homemade options.

ItSpawnedaZillionTrends.Or,atLeast,“Trends”

Including but certainly not limited to: Butter boards, sexy water, blueberry milk nails, unexpected red, lucky girl syndrome, first-time-cool syndrome, bed rotting, 75 soft, 75 cozy, bookshelf wealth, loud budgeting, broccoli freckles, strawberry makeup, glazed donut skin, latte makeup, cowboy copper hair, old money blonde, expensive brunette, orange peel theory, quiet luxury, stealth wealth, tomato girl summer,  indie sleaze, coquette, looksmaxxing, male perms, vanilla girl, clean girl, soft girl, coastal grandma, coastal cowgirl, low-high visual weight makeup, sleepy-girl mocktails, fluffy coffee, shrimp tree, girl math, girl dinner, mob wife, clowncore, balletcore, Barbiecore, royalcore, corecore.

ItGotTaylorSwifttoDefyHerLabel

For the music industry, TikTok has become a potent but unpredictable promotional outlet, and a vital one in the race to mint a new hit. Young artists like Olivia Rodrigo and Lil Nas X saw their popularity explode on the platform, and acts like Fleetwood Mac have seen decades-old songs get a boost from memes on the app.

But TikTok is also the latest tech platform to draw the anger of the music industry for low royalty rates. In February, Universal Music Group, which represents artists like Ms. Rodrigo, Taylor Swift, Billie Eilish and Drake, withdrew the rights to its music on the app, saying that TikTok was trying to “bully” the company to accept low terms.

Within days, millions of TikTok videos using music from Universal artists went mute, and since then guessing which side would blink first has become a media-business parlor game.

Last week, however, Ms. Swift — who releases her music through Universal, but has owned the copyrights to her work since 2018 — broke ranks and put her songs back on TikTok, just ahead of the release of her next album on Friday. Now the question is, will other artists will follow.

ItMightJustBeaWeapon

For years I thought TikTok was mostly a parenting problem, and had only tangential bearing on what I cover: threats to national security. It took a while — and a lot of conversations with both tech firms and government officials — for me to become concerned about the potential that it could also pose a major problem on that front.

Not because the company’s Chinese owners could figure out your dance-move preferences, but because the algorithm at the core of the app is wrapped in such mystery.

So what’s the issue? The algorithm doesn’t belong to TikTok; it is provided by engineers working for ByteDance, the Chinese company that controls the platform and develops the code in enormous secrecy in laboratories around the world, in Beijing, Singapore and Mountain View, Calif.

No one outside the company knows exactly what goes into those algorithms.

The Chinese government is intent on keeping it that way. It has issued regulations that require Beijing’s regulators to grant permission before any ByteDance algorithms can be licensed to outsiders. They are unlikely to do so.

And so, as long as it is written by ByteDance, and can’t be picked apart on the outside, there will always be the risk that it will become a pipeline for influencing citizens, and thus voters, in subtle and not-so-subtle ways.

Senator Mark Warner, the chairman of Senate Intelligence Committee, has noted that because TikTok has emerged as a major source of news — and because it collects data on users that the Chinese government could find useful, even crucial — it poses a serious threat, and could become “the most powerful propaganda tool ever.”

Of course, that threat is mostly hypothetical at this point.

At least based on what the United States has made public. The intelligence agencies have been giving closed-door briefings, but presumably there hasn’t been a classified blockbuster, since there would likely be great pressure to declassify it.

Still, we have seen waves of new influence campaigns flowing out of China — much of it aimed at nations other than the United States. While TikTok has not been at the center of those campaigns, clearly, the Chinese have learned a lot in the past few years, including from the Russians. (Researchers have also found that topics commonly suppressed in China, including about the Tibetan and Uyghur populations, appear to be unusually underrepresented on TikTok compared with Instagram.)

This is not a problem that would be solved by simply selling TikTok’s operations to an American buyer. Sure, the bill that went through the House bans a new, Western-owned TikTok from having any “operational relationship” with ByteDance, “including any cooperation with respect to the operation of a content recommendation algorithm.” Good luck with that — TikTok would no longer be TikTok.

The real question is whether anyone gets to look under the hood. Because to make Americans trust TikTok, the country will need an early warning system, something that will assure everyone that a technology that became popular because it generated memes and celebrates self-expression does not become a conduit for a foreign government interested in subtly influencing how we vote.

ItTurnedtheSchoolBathroomIntoaMovieSet

Southern Alamance Middle School, a public school in Graham, N.C., recently came up with a novel way to combat student distractions from social media. Or at least to curb the phenomenon that some teachers have dubbed “Toilet TikToks.”

The problem: Educators there noticed a spike in the number of students asking to leave class — sometimes as frequently as nine times per day — to go to the bathroom, where they made TikTok videos.

The solution: Administrators decided to remove the bathroom mirrors that students used to film TikToks and primp for their close-ups. They also introduced an online system that issues students digital hall passes when they want to be excused from class and that allows administrators to track students’ locations. “Since removing the mirrors,” administrators wrote in a message to parents in January, “we have seen a drastic decrease in bathroom visits from students asking to be excused just to make videos.”

Toilettoks — a TikTok genre, dating back at least five years, in which students use school bathrooms as film sets for dance routines, lip-syncing clips or critiques of unclean lavatories — are one of the milder social media annoyances for schools.

Across the United States, students have also used school bathrooms as arenas to stage, film and post videos of bullying, physical assaults on schoolmates and acts of vandalism.

In March, Alamance-Burlington schools announced that it was joining dozens of other U.S. districts that have filed lawsuits accusing social media platforms, including TikTok, of unfairly ensnaring young people.

“We’re seeing the negative impacts of social media on our students every day,” Kristy Davis, the acting superintendent of Alamance-Burlington schools, said. “Their well-being has to be the top priority.”

ItTookOverAlltheOtherApps

My favorite Instagram account is a collection of TikToks. Curated by the videographer Leia Jospé, @favetiktoks420 hunts for Gen-Z’s ickiest thirst traps and bleakest acting exercises and delivers them to me in a Millennial-safe package, uploading them directly to a social network that I actually use.

By the time TikTok debuted, in 2017, I was already in my 30s and too old and lazy to work another app into my rotation. Instagram and Twitter were distracting enough. But now those platforms lie downstream of TikTok’s creative wellspring, waiting for bits of its most popular content to drift into the open internet. TikToks float into my friends’ Instagram stories, percolate into our group chats, swirl into my Twitter feed. My phone is always bleating with its outro sound effect. I rarely open TikTok, but I watch TikToks all the time.

TikToks let loose a chaotic element into Instagram’s internet mall, and they break the monotony of Twitter’s boosted tech-bro threads. They stock YouTube compilations and spark Facebook debates and fuel trend pieces. 

If TikTok were to disappear, it would feel, at least for a while, like the internet’s big content spigot had been turned to a trickle. Rival platforms have tried to remake themselves in TikTok’s image — building in short-form videos, algorithmic timelines and searchable sound clips — but have failed to reproduce the hypnotic energy of its perpetual discovery machine. We’d be left with a diluted version of its secret sauce.

But any network that hopes to capitalize on its own popularity will disrupt its product. Even as other social media platforms try to become TikTok, TikTok is trying to become them, lengthening its videos to compete with YouTube and introducing an e-commerce platform to “drive meaningful shopping experiences” and rival Instagram. Eventually some new, inexplicably addictive platform will rise in its place. And I will rely on the kindness of some slightly younger strangers to show me what’s on it.

ItMimicsaFlowState

Much has been said about the “addictive design” of TikTok. But what is the social media site actually doing to our brains?

There is very little research looking at what goes on inside people’s heads while they’re using TikTok. But one small study conducted on Chinese university students used magnetic resonance imaging to compare brain activity while they watched personalized TikTok videos (ones the algorithm had selected based on their past use) versus generalized ones (videos the app recommended to new users).

The students had greater activity in several areas of the brain, including ones associated with reward, attention and processing social information, while viewing personalized videos. In other words, the algorithm did its job.

Other social media platforms have been shown to turn on similar brain regions. So what makes TikTok different? Some experts have proposed that it can send users into a “flow state”: the experience of being so absorbed in a task that the person loses track of time. Backing this up, one study found that TikTok users reported experiencing higher levels of flow than Instagram users.

“Flow” is often associated with work or hobbies — activities that are challenging enough to be engaging but not frustrating. Watching videos doesn’t require skill the way that many flow-inducing activities do, yet the app is able to induce the feelings of enjoyment, concentration and time distortion that are characteristic of flow — possibly because of the algorithm’s immersive quality.

ItWonOvertheSnobbiestGatekeeper

Is there any more official signal that a business titan has arrived at the heart of the American social-financial-artistic-political power nexus than being invited to be an honorary host of the annual Met Gala, a.k.a. “the party of the year”? Any more glamorous recognition than being asked to join its convener, the Vogue editor, Anna Wintour, in the Metropolitan Museum’s soaring atrium as the great and the good of Hollywood, fashion, sports, Wall Street and Washington swan past? 

On May 6, TikTok will be lead sponsor of both the party and the museum fashion exhibition it celebrates. The company’s chief executive, Shou Chew, has been named an honorary chair of this year’s gala, along with the Loewe designer Jonathan Anderson, while Ms. Wintour, Zendaya, Bad Bunny, Jennifer Lopez and Chris Hemsworth are the event’s co-chairs.

That placement would put TikTok firmly in the tradition of previous gala sponsors like Amazon, Instagram and Apple — tech companies bedazzled by the Old Establishment, which in turn is bedazzled by their blush of upstart cool. 

It is an acknowledgment, if any were needed, of the prominent role the app has come to play in fashion in a mere few years.

It was only in 2021, after all, that Ms. Wintour was criticized for inviting TikTok stars such as Addison Rae and Dixie D’Amelio to the party — for somehow cheapening it by catering to the buzzfeed machine of the smartphone, rather than the elite. After all, not just anyone can get an invite, even if they can afford the $50,000 price tag for a seat; Ms. Wintour vets every guest, and the price of admission has to do with cultural currency even more than actual currency.

Which is why, of course, TikTok belongs. Despite the fact that all social media is forbidden inside the party.

Fast-forward three years, and there are more than 75 billion views associated with the #TikTokfashion hashtag; almost 500 million with #2023Gala alone. Luxury brands routinely sign up TikTok stars as brand ambassadors along with every other kind of star, hoping to access their audience (received relevance is something Vogue might be getting out of the association, too). And thus is created a virtuous — or vicious? — cycle in which TikTok feeds the gala machine, which feeds TikTok, which is the vicarious experience that has come to feed us all.

ItLedPeopletoSelf-DiagnoseADHD

TikTok is a mother lode of mental health content, filled with compelling first-person accounts of everything from major depression to selective mutism. Depending on your perspective, that’s either a very good thing — or concerning.

Corey Basch, who analyzed 100 popular TikTok videos with the hashtag #mentalhealth for a 2022 study, emerged concerned about the looping effect of the algorithm. 

“What’s so important and disturbing to recognize is the downward spiral that users can get swept into,” said Dr. Basch, a professor of public health at William Paterson University. “If one is drawn to posts related to despair and anxiety, they can easily spend hours exposed to repetitive content known as an echo chamber.”

The surge of content about mental health has meant that young people are more likely to self-diagnose before seeing a clinician, psychiatrists report. Diagnoses for ADHD and anxiety disorders shot up during the pandemic years, especially among young people. 

Some researchers have expressed concern about how profit motives may feed into these trends, since platforms often feature advertising from app-based mental health services, and influencers have sponsorship deals with such companies.

“They say we can diagnose you really quick, just take this five-question quiz and we can send you a prescription in a nice little box,” said Holly Avella, a Ph.D. candidate at Rutgers University who has researched mental health and social media. 

Researchers also warn that TikTok videos can deliver misinformation. A review of literature published last year found that around one-fifth of videos mentioning cognitive behavioral therapy were inaccurate, describing it as ineffective or harmful. 

But some users credit the app with breaking open the national conversation around mental illness.

“You can sit there on your pedestal and pooh-pooh it all you want,” said Kate Speer, who has used her social media feeds to describe her experience of serious mental illness. TikTok is helpful for “the very people who are struggling the most, those who don’t have access to services in the real world and who might even be so disabled by mental illness that they are locked in their houses.”

It’stheLatestCampaignTool

President Biden turned down an opportunity to appear on CBS and reach tens of millions of potential voters tuning in for this year’s Super Bowl. Instead, he released his first TikTok.

“Chiefs or Niners?” asked a disembodied, youthful-sounding voice. “Two great quarterbacks; hard to decide,” replied the president, casually dressed in a half-zip sweater and khakis. The caption was “lol hey guys.”

Team Biden, like most other major politicians, had previously resisted joining the app because of security concerns related to its Chinese ownership. (The Donald Trump campaign is not on TikTok, and Mr. Trump has expressed divergent views about the app, proposing a ban during his presidency but recently criticizing an attempt by Congress to curtail its use in the United States.)

Giving in was a nod to the irrefutable importance of TikTok, where about 14 percent of American adults regularly get news, in an election year. There’s now a small studio in the Biden campaign office in Wilmington, Del., where staff members can film “candid” videos with the candidate.

62% of Americans between 18- and 29-years-old use the platform, greater than the share of that age group that voted in the last presidential election.

Campaigns have a rich tradition of adapting to the latest technological fads, from wireless radio to television sets and, more recently, to social platforms like Facebook and Snapchat. Many of these efforts share a how-do-you-do-fellow-kids quality to them, and in an attempt to avoid appearing out of touch, the Biden campaign relies on young, digitally fluent aides to host its TikToks.

It works, sometimes. One video claims to have Trump “caught on camera” making offensive remarks, an attempt to replicate the amateur spontaneity of many TikToks. (In reality, it’s someone’s iPhone aimed at a TV broadcast of a Trump speech.) Other times it comes across try-hard-y, like the video that dismisses a post by Representative Jim Jordan using a popular “I Ain’t Reading All That” online meme.

The @bidenhq account, though, is hovering around 299,000 followers — still small beans in the TikTok world. But in a close race, every lol counts.

It’stheNewMall

Most months, when Kiara Springs posts on her TikTok accountabout mini skirts or linen tops she finds on Amazon, she earns $10,000 to $12,000 for getting people to buy what she suggests. During her biggest month, Ms. Springs, 25, raked in $50,000 for her posts.

TikTok is now a multibillion-dollar shopping experience — and companies have glommed on. The internet might have killed malls, but now it is one big mall. 

Because the bite-size videos are addictive, and partly because advertising on the platform is relatively inexpensive for smaller brands, the app has become a core part of many companies’ marketing plans. Brands say that their videos populated with everyday people can more easily go viral than on, say, Instagram, where they often need to pay expensive influencers. And people who notice shopping-related content spend more time on TikTok, according to eMarketer.

The average user spends nearly an hour — 58 minutes — per day on the platform.

Last year, TikTok debuted a prominent shopping feed on the app that now allows people to buy goods directly from a wide array of vendors. Some fashion and beauty brands think about the TikTok content they could make for a product before developing it.

Fiona Co Chan, a co-founder of Youthforia, a beauty and skin care brand with roughly 190,000 followers on the app, says if she can’t think of 200 TikTok videos that she could make for a product, she’ll likely scrap it entirely.

ItMadeYouBuyThatOneWaterBottle

When a product goes viral on TikTok, those views often translate directly into increased sales. In some cases, the effect has been dramatic:

  • Stanley tumblersCompany revenue last year was $750 million, up from $73 million in 2019, after the product became a sensation on TikTok.

  • Feta cheeseDemand jumped 200 percent at one grocery chain in 2021, after a recipe for baked feta pasta took off and amassed more than 20 million views.

  • CeraveSales increased by more than 60 percent in 2020 after skin care became a lockdown pastime and TikTok users discovered the drugstore mainstay.

  • Cat Crack Catnip It briefly sold out in 2021 after TikTok users posted videos of their cats going crazy for it.

  • Isle of Paradise tanning sprayIt sold out in 48 hours in 2021 after a post about it went viral.

  • Prepdeck kitchen storage productsThey went out of stock in 2021 after going viral.

ItTurnedRecipesIntoConcepts

Cooking is different now. One crucial distinction between the hit recipes of today (like Emily Mariko’s salmon and rice bowl) and those of yesteryear (The Silver Palate’s chicken marbella) is the medium.

Before, a static image was all you had to get the point across in a recipe, with step-by-step instructions printed in a cookbook.

Now recipes unfold over time. In a 30-second video, there are obvious visual cues that viewers can absorb, techniques they can sink their teeth into. The videos depict process, not just stages, and allow you to jump-cut your way through a recipe in a few blinks.

But for all the access to techniques and cuisines that TikTok has provided home cooks, the platform favors concepts, over actual recipes — eggs fried in a puddle of pesto, sandwich fillings chopped into a homogenous mixture, mini pancakes served like cereal. The most shareable recipes are the ones that you can watch once, then turn around and make — no measurements, bake times or reading needed. Just dump, stir, like, follow, repeat.

ItKnowsWhereYouLive

Every social media app is, essentially, a spy in your pocket. When it comes to data collection, TikTok is no worse than the others. The main difference, and the one that’s driving the current conversation in Washington, is that TikTok is owned by a Chinese company.

TikTok, like other apps, seeks a huge amount of information from you, and some pieces of it that can seem innocuous are quite revealing, including:

Your I.P. address and location

An I.P. address is a unique identifier associated with your device or the network you use to get on the internet. TikTok can use your I.P. address and location to determine the advertising you see, but it can also reveal other real-life associations. When people access a social network from the same I.P. address, it reveals that they may know one another offline. ByteDance, which owns TikTok, used I.P. address data collected from journalists using the app to try to identify company employees who were speaking to them.

Your contacts

Giving TikTok access to the hundreds (or thousands) of numbers and email addresses on your phone — an opt-in feature — lets them draw unexpected insights into your life, such as who your doctors are, your present and former colleagues, your one-night stands, and on and on. TikTok may then recommend you follow them as “people you may know” — and your account, in turn, will be suggested to them, so beware of sharing contacts if you’re trying to stay anonymous. Even if you don’t share your contacts, TikTok can look for the phone or email address you gave the company in other users’ address books unless you go into the privacy settings and turn off “Suggest Your Account to Others.”

Your messages

Unless your messages are end-to-end encrypted, which they are not on TikTok, they can be reviewed by the company storing them for you.

Your viewing history

Are you obsessed with steamy Twilight highlight reels or home repair videos? TikTok knows.

ItTeaches22-Year-OldsHowtoLiveTheirBestMiddle-AgeLife

Shelley Polanco is 22, and a senior at Brandeis. As she faces life after college, she is not looking for practical advice about, say, jobs and careers. She yearns instead for a guidebook to what it feels like to be an adult, something to tell her about her future emotional life and satisfaction.

And so, she likes following TikTok accounts that feature “an older woman of culture, kind of like this auntie figure who gets on TikTok and records ‘things I wish I would have known in my 20s.’”

She’s a fan of @itsrealllylola. “She’s turning 25, and she starts to speak about all the things she’s learned, the ways you maybe want to ignore people’s judgments on you,” Ms. Polanco says. “She’s big on ‘live your life and have fun.’”

She peers decades into her future with the help of Dr. Amanda Hanson, “#midlifemuse,” who invites comments about life after 50 from her TikTok followers. Or  Shera Seven. “She is this 40-year-old woman, I want to say, giving really brutal dating advice,” Ms. Polanco said. 

Ms. Polanco was having trouble seeing more than three months out, and found support from a TikTok transformation coach named Shannon. “I searched, ‘how to write goals,’” she said. “There was this lady walking you through this meditation to visualize your future self, and it was one of the only times I could see an older version of myself. It was so inspirational, I closed the app, got out a piece of paper and started writing.”

ItHelpedYouAskforaRaise

In a video by the account @salarytransparentstreet that’s been viewed more than 23 million times, a lawyer candidly shares she makes $134,000 a year, a teacher says she earns $53,000 and a man who does chemical risk assessments for the federal government divulges he makes $60,000 a year. The point, according to Hannah Williams, the 27-year-old content creator behind the account, is to help people better understand what they could be earning. (Ms. Williams says she made more than $1 million in 2023 before expenses through her videos and earns revenue from brand sponsorships and ads).

FinTok, as the money and personal finance community on the app is called, has fundamentally changed the way we accept advice from strangers — and altered how much ordinary people are willing to share about that most taboo subject: how much money they have.

Sure, the app has allowed established finance gurus like Dave Ramsey and Suze Orman to expand their empires. But is has also given rise to people like Vivian Tu, a 30-year-old former Wall Street trader who runs the account @yourrichbff. Ms. Tu, who has 2.5 million followers, offers practical advice on high-yield savings accounts and retirement savings, but can also take widely discussed TikTok drama and show people what financial lessons they can learn from it.

57% of Gen Z users like or leave a comment after watching a video on the platform.

As with anything on TikTok, scrolling through FinTok videos requires a certain amount of skepticism. There are plenty of cryptocurrency creators on TikTok who like to focus on the potential gains rather than warning people of the risks. There are also crypto scams, including a rampant one using deepfake videos of Elon Musk. (Ads for crypto or financial services are bannedfrom the platform, in part to help protect people from getting involved with high-risk investments. )

But often, when scams or bad advice crop up, commenters have no problem offering corrections. When one creator posted a videoabout how to get a high credit limit using dubious methods, her comment section quickly filled up with users accusing her of committing fraud. The video was soon deleted.

ItTiltedYourScreen

TikTok didn’t invent vertically oriented videos. But it has been very influential in getting people to watch their screens upright instead of sideways. It’s a phenomenon that is sticking elsewhere, with Apple, a professional Spanish soccer league and major news publishers all producing vertical videos. Even The New York Times is on board.

Design and development by Michael Beswetherick

Editing by Noreen Malone, Ashwin Seshagiri, Matt Ruby and Sharon O’Neal

Additional production by Brent Murray, Amanda Cordero and Joshua Shao


How TikTok Changed America - The New York Times

Saturday, April 06, 2024

How Tech Giants Cut Corners to Harvest Data for A.I. - The New York Times

How Tech Giants Cut Corners to Harvest Data for A.I.

"OpenAI, Google and Meta ignored corporate policies, altered their own rules and discussed skirting copyright law as they sought online information to train their newest artificial intelligence systems.

Researchers at OpenAI’s office in San Francisco developed a tool to transcribe YouTube videos to amass conversational text for A.I. development. Jason Henry for The New York Times

In late 2021, OpenAI faced a supply problem.

The artificial intelligence lab had exhausted every reservoir of reputable English-language text on the internet as it developed its latest A.I. system. It needed more data to train the next version of its technology — lots more.

So OpenAI researchers created a speech recognition tool called Whisper. It could transcribe the audio from YouTube videos, yielding new conversational text that would make an A.I. system smarter.

Some OpenAI employees discussed how such a move might go against YouTube’s rules, three people with knowledge of the conversations said. YouTube, which is owned by Google, prohibits use of its videos for applications that are “independent” of the video platform.

Ultimately, an OpenAI team transcribed more than one million hours of YouTube videos, the people said. The team included Greg Brockman, OpenAI’s president, who personally helped collect the videos, two of the people said. The texts were then fed into a system called GPT-4, which was widely considered one of the world’s most powerful A.I. models and was the basis of the latest version of the ChatGPT chatbot.

The race to lead A.I. has become a desperate hunt for the digital data needed to advance the technology. To obtain that data, tech companies including OpenAI, Google and Meta have cut corners, ignored corporate policies and debated bending the law, according to an examination by The New York Times.

At Meta, which owns Facebook and Instagram, managers, lawyers and engineers last year discussed buying the publishing house Simon & Schuster to procure long works, according to recordings of internal meetings obtained by The Times. They also conferred on gathering copyrighted data from across the internet, even if that meant facing lawsuits. Negotiating licenses with publishers, artists, musicians and the news industry would take too long, they said.

Like OpenAI, Google transcribed YouTube videos to harvest text for its A.I. models, five people with knowledge of the company’s practices said. That potentially violated the copyrights to the videos, which belong to their creators.

Last year, Google also broadened its terms of service. One motivation for the change, according to members of the company’s privacy team and an internal message viewed by The Times, was to allow Google to be able to tap publicly available Google Docs, restaurant reviews on Google Maps and other online material for more of its A.I. products.

The companies’ actions illustrate how online information — news stories, fictional works, message board posts, Wikipedia articles, computer programs, photos, podcasts and movie clips — has increasingly become the lifeblood of the booming A.I. industry. Creating innovative systems depends on having enough data to teach the technologies to instantly produce text, images, sounds and videos that resemble what a human creates.

The volume of data is crucial. Leading chatbot systems have learned from pools of digital text spanning as many as three trillion words, or roughly twice the number of words stored in Oxford University’s Bodleian Library, which has collected manuscripts since 1602. The most prized data, A.I. researchers said, is high-quality information, such as published books and articles, which have been carefully written and edited by professionals.

For years, the internet — with sites like Wikipedia and Reddit — was a seemingly endless source of data. But as A.I. advanced, tech companies sought more repositories. Google and Meta, which have billions of users who produce search queries and social media posts every day, were largely limited by privacy laws and their own policies from drawing on much of that content for A.I.

Their situation is urgent. Tech companies could run through the high-quality data on the internet as soon as 2026, according to Epoch, a research institute. The companies are using the data faster than it is being produced.

“The only practical way for these tools to exist is if they can be trained on massive amounts of data without having to license that data,” Sy Damle, a lawyer who represents Andreessen Horowitz, a Silicon Valley venture capital firm, said of A.I. models last year in a public discussion about copyright law. “The data needed is so massive that even collective licensing really can’t work.”

Tech companies are so hungry for new data that some are developing “synthetic” information. This is not organic data created by humans, but text, images and code that A.I. models produce — in other words, the systems learn from what they themselves generate.

OpenAI said each of its A.I. models “has a unique data set that we curate to help their understanding of the world and remain globally competitive in research.” Google said that its A.I. models “are trained on some YouTube content,” which was allowed under agreements with YouTube creators, and that the company did not use data from office apps outside of an experimental program. Meta said it had “made aggressive investments” to integrate A.I. into its services and had billions of publicly shared images and videos from Instagram and Facebook for training its models.

For creators, the growing use of their works by A.I. companies has prompted lawsuits over copyright and licensing. The Times sued OpenAI and Microsoft last year for using copyrighted news articles without permission to train A.I. chatbots. OpenAI and Microsoft have said using the articles was “fair use,” or allowed under copyright law, because they transformed the works for a different purpose.

More than 10,000 trade groups, authors, companies and others submitted comments last year about the use of creative works by A.I. models to the Copyright Office, a federal agency that is preparing guidance on how copyright law applies in the A.I. era.

Justine Bateman, a filmmaker, former actress and author of two books, told the Copyright Office that A.I. models were taking content — including her writing and films — without permission or payment.

“This is the largest theft in the United States, period,” she said in an interview.

‘Scale Is All You Need’

Jared Kaplan, a theoretical physicist at Johns Hopkins University, wrote a key paper on A.I. and data. He is also the chief science officer of the A.I. start-up Anthropic. Chris J. Ratcliffe/Bloomberg

In January 2020, Jared Kaplan, a theoretical physicist at Johns Hopkins University, published a groundbreaking paper on A.I. that stoked the appetite for online data.

His conclusion was unequivocal: The more data there was to train a large language model — the technology that drives online chatbots — the better it would perform. Just as a student learns more by reading more books, large language models can better pinpoint patterns in text and be more accurate with more information.

“Everyone was very surprised that these trends — these scaling laws as we call them — were basically as precise as what you see in astronomy or physics,” said Dr. Kaplan, who published the paper with nine OpenAI researchers. (He now works at the A.I. start-up Anthropic.)

“Scale is all you need” soon became a rallying cry for A.I.

Researchers have long used large public databases of digital information to develop A.I., including Wikipedia and Common Crawl, a database of more than 250 billion web pages collected since 2007. Researchers often “cleaned” the data by removing hate speech and other unwanted text before using it to train A.I. models.

In 2020, data sets were tiny by today’s standards. One database containing 30,000 photographs from the photo website Flickr was considered a vital resource at the time.

After Dr. Kaplan’s paper, that amount of data was no longer enough. It became all about “just making things really big,” said Brandon Duderstadt, the chief executive of Nomic, an A.I. company in New York.

When OpenAI unveiled GPT-3 in November 2020, it was trained on the largest amount of data to date — about 300 billion “tokens,” which are essentially words or pieces of words. After learning from that data, the system generated text with astounding accuracy, writing blog posts, poetry and its own computer programs.

In 2022, DeepMind, an A.I. lab owned by Google, went further. It tested 400 A.I. models and varied the amount of training data and other factors. The top-performing models used even more data than Dr. Kaplan had predicted in his paper. One model, Chinchilla, was trained on 1.4 trillion tokens.

It was soon overtaken. Last year, researchers from China released an A.I. model, Skywork, which was trained on 3.2 trillion tokens from English and Chinese texts. Google also unveiled an A.I. system, PaLM 2, which topped 3.6 trillion tokens.

Transcribing YouTube

In May, Sam Altman, the chief executive of OpenAI, acknowledged that A.I. companies would use up all viable data on the internet.

“That will run out,” he said in a speech at a tech conference.

Mr. Altman had seen the phenomenon up close. At OpenAI, researchers had gathered data for years, cleaned it and fed it into a vast pool of text to train the company’s language models. They had mined the computer code repository GitHub, vacuumed up databases of chess moves and drawn on data describing high school tests and homework assignments from the website Quizlet.

By late 2021, those supplies were depleted, said eight people with knowledge of the company, who were not authorized to speak publicly.

OpenAI was desperate for more data to develop its next-generation A.I. model, GPT-4. So employees discussed transcribing podcasts, audiobooks and YouTube videos, the people said. They talked about creating data from scratch with A.I. systems. They also considered buying start-ups that had collected large amounts of digital data.

OpenAI eventually made Whisper, the speech recognition tool, to transcribe YouTube videos and podcasts, six people said. But YouTube prohibits people from not only using its videos for “independent” applications, but also accessing its videos by “any automated means (such as robots, botnets or scrapers).”

OpenAI employees knew they were wading into a legal gray area, the people said, but believed that training A.I. with the videos was fair use. Mr. Brockman, OpenAI’s president, was listed in a research paper as a creator of Whisper. He personally helped gather YouTube videos and fed them into the technology, two people said.

Mr. Brockman referred requests for comment to OpenAI, which said it uses “numerous sources” of data.

Last year, OpenAI released GPT-4, which drew on the more than one million hours of YouTube videos that Whisper had transcribed. Mr. Brockman led the team that developed GPT-4.

Some Google employees were aware that OpenAI had harvested YouTube videos for data, two people with knowledge of the companies said. But they didn’t stop OpenAI because Google had also used transcripts of YouTube videos to train its A.I. models, the people said. That practice may have violated the copyrights of YouTube creators. So if Google made a fuss about OpenAI, there might be a public outcry against its own methods, the people said.

Matt Bryant, a Google spokesman, said the company had no knowledge of OpenAI’s practices and prohibited “unauthorized scraping or downloading of YouTube content.” Google takes action when it has a clear legal or technical basis to do so, he said.

Google’s rules allowed it to tap YouTube user data to develop new features for the video platform. But it was unclear whether Google could use YouTube data to build a commercial service beyond the video platform, such as a chatbot.

Geoffrey Lottenberg, an intellectual property lawyer with the law firm Berger Singerman, said Google’s language about what it could and could not do with YouTube video transcripts was vague.

“Whether the data could be used for a new commercial service is open to interpretation and could be litigated,” he said.

In late 2022, after OpenAI released ChatGPT and set off an industrywide race to catch up, Google researchers and engineers discussed tapping other user data. Billions of words sat in people’s Google Docs and other free Google apps. But the company’s privacy restrictions limited how they could use the data, three people with knowledge of Google’s practices said.

After OpenAI released ChatGPT, Google researchers and engineers discussed tapping other user data to develop A.I. products, people with knowledge of the discussions said. Jason Henry for The New York Times

In June, Google’s legal department asked the privacy team to draft language to broaden what the company could use consumer data for, according to two members of the privacy team and an internal message viewed by The Times.

The employees were told Google wanted to use people’s publicly available content in Google Docs, Google Sheets and related apps for an array of A.I. products. The employees said they didn’t know if the company had previously trained A.I. on such data.

At the time, Google’s privacy policy said the company could use publicly available information only to “help train Google’s language models and build features like Google Translate.”

The privacy team wrote new terms so Google could tap the data for its “A.I. models and build products and features like Google Translate, Bard and Cloud AI capabilities,” which was a wider collection of A.I. technologies.

“What is the end goal here?” one member of the privacy team asked in an internal message. “How broad are we going?”

The team was told specifically to release the new terms on the Fourth of July weekend, when people were typically focused on the holiday, the employees said. The revised policy debuted on July 1, at the start of the long weekend.

How Google Can Use Your Data

Here are the changes Google made to its privacy policy last year for its free consumer apps.

Google uses information to improve our services and to develop new products, features and technologies that benefit our users and the public. For example, we use publicly available information to help train Google’s language AI models and build products and features like Google Translate, Bard, and Cloud AI capabilities.

In August, two privacy team members said, they pressed managers on whether Google could start using data from free consumer versions of Google Docs, Google Sheets and Google Slides. They were not given clear answers, they said.

Mr. Bryant said that the privacy policy changes had been made for clarity and that Google did not use information from Google Docs or related apps to train language models “without explicit permission” from users, referring to a voluntary program that allows users to test experimental features.

“We did not start training on additional types of data based on this language change,” he said.

The Debate at Meta

Mark Zuckerberg, Meta’s chief executive, had invested in A.I. for years — but suddenly found himself behind when OpenAI released ChatGPT in 2022. He immediately pushed to match and exceed ChatGPT, calling executives and engineers at all hours of the night to push them to develop a rival chatbot, said three current and former employees, who were not authorized to discuss confidential conversations.

But by early last year, Meta had hit the same hurdle as its rivals: not enough data.

Ahmad Al-Dahle, Meta’s vice president of generative A.I., told executives that his team had used almost every available English-language book, essay, poem and news article on the internet to develop a model, according to recordings of internal meetings, which were shared by an employee. 

Meta could not match ChatGPT unless it got more data, Mr. Al-Dahle told colleagues. In March and April 2023, some of the company’s business development leaders, engineers and lawyers met nearly daily to tackle the problem.

Some debated paying $10 a book for the full licensing rights to new titles. They discussed buying Simon & Schuster, which publishes authors such as J.K. Rowling and Stephen King, according to the recordings.

They also talked about how they had summarized books, essays and other works from the internet without permission and discussed sucking up more, even if that meant facing lawsuits. One lawyer warned of “ethical” concerns around taking intellectual property from artists but was met with silence, according to the recordings.

Mr. Zuckerberg demanded a solution, employees said.

“The capability that Mark is looking for in the product is just something that we currently aren’t able to deliver,” one engineer said.

Mark Zuckerberg, Meta’s chief executive, pushed his company to catch up in generative A.I. after OpenAI released ChatGPT.Jason Andrew for The New York Times

While Meta operates giant social networks, it didn’t have troves of user posts at its disposal, two employees said. Many Facebook users had deleted their earlier posts, and the platform wasn’t where people wrote essay-type content, they said.

Meta was also limited by privacy changes it introduced after a 2018 scandal over sharing its users’ data with Cambridge Analytica, a voter-profiling company.

Mr. Zuckerberg said in a recent investor call that the billions of publicly shared videos and photos on Facebook and Instagram are “greater than the Common Crawl data set.”

During their recorded discussions, Meta executives talked about how they had hired contractors in Africa to aggregate summaries of fiction and nonfiction. The summaries included copyrighted content “because we have no way of not collecting that,” a manager said in one meeting.

Meta’s executives said OpenAI seemed to have used copyrighted material without permission. It would take Meta too long to negotiate licenses with publishers, artists, musicians and the news industry, they said, according to the recordings.

“The only thing that’s holding us back from being as good as ChatGPT is literally just data volume,” Nick Grudin, a vice president of global partnership and content, said in one meeting.

OpenAI appeared to be taking copyrighted material and Meta could follow this “market precedent,” he added.

Meta’s executives agreed to lean on a 2015 court decision involving the Authors Guild versus Google, according to the recordings. In that case, Google was permitted to scan, digitize and catalog books in an online database after arguing that it had reproduced only snippets of the works online and had transformed the originals, which made it fair use.

Using data to train A.I. systems, Meta’s lawyers said in their meetings, should similarly be fair use.

At least two employees raised concerns about using intellectual property and not paying authors and other artists fairly or at all, according to the recordings. One employee recounted a separate discussion about copyrighted data with senior executives including Chris Cox, Meta’s chief product officer, and said no one in that meeting considered the ethics of using people’s creative works.

‘Synthetic’ Data

OpenAI’s Mr. Altman had a plan to deal with the looming data shortage.

Companies like his, he said at the May conference, would eventually train their A.I. on text generated by A.I. — otherwise known as synthetic data.

Since an A.I. model can produce humanlike text, Mr. Altman and others have argued, the systems can create additional data to develop better versions of themselves. This would help developers build increasingly powerful technology and reduce their dependence on copyrighted data.

“As long as you can get over the synthetic data event horizon, where the model is smart enough to make good synthetic data, everything will be fine,” Mr. Altman said.

A.I. researchers have explored synthetic data for years. But building an A.I system that can train itself is easier said than done. A.I. models that learn from their own outputs can get caught in a loop where they reinforce their own quirks, mistakes and limitations.

“The data these systems need is like a path through the jungle,” said Jeff Clune, a former OpenAI researcher who now teaches computer science at the University of British Columbia. “If they only train on synthetic data, they can get lost in the jungle.”

To combat this, OpenAI and others are investigating how two different A.I. models might work together to generate synthetic data that is more useful and reliable. One system produces the data, while a second judges the information to separate the good from the bad. Researchers are divided on whether this method will work.

A.I. executives are barreling ahead nonetheless.

“It should be all right,” Mr. Altman said at the conference.

Cade Metz writes about artificial intelligence, driverless cars, robotics, virtual reality and other emerging areas of technology. More about Cade Metz

Sheera Frenkel is a reporter based in the San Francisco Bay Area, covering the ways technology impacts everyday lives with a focus on social media companies, including Facebook, Instagram, Twitter, TikTok, YouTube, Telegram and WhatsApp. More about Sheera Frenkel

Nico Grant is a technology reporter covering Google from San Francisco. Previously, he spent five years at Bloomberg News, where he focused on Google and cloud computing."

How Tech Giants Cut Corners to Harvest Data for A.I. - The New York Times