Mistral

French AI firm Mistral to build data centres in Sweden

  • The 1.2 billion euro ($1.4 billion) investment is "a major step toward Europe's technological independence", Mistral said, offering "a completely European AI solution".
  • French AI developer Mistral said Wednesday that it would build data centres in Sweden, its first outside France, as it races to compete with the sector's biggest names.
  • The 1.2 billion euro ($1.4 billion) investment is "a major step toward Europe's technological independence", Mistral said, offering "a completely European AI solution".
French AI developer Mistral said Wednesday that it would build data centres in Sweden, its first outside France, as it races to compete with the sector's biggest names.
The 1.2 billion euro ($1.4 billion) investment is "a major step toward Europe's technological independence", Mistral said, offering "a completely European AI solution".
One of Europe's leading lights in artificial intelligence, Mistral has posted revenues far behind those of American competitors such as OpenAI and Anthropic, though profitability remains elusive for most of the sector.
But Mistral's European DNA may prove to be an advantage, with technological sovereignty increasingly on leaders' minds.
It has focussed in particular on business clients and applications, and not chatbots like ChatGPT or Claude that target everyday users.
In September it raised 1.7 billion euros, bringing aboard Dutch chipmaking technology giant ASML as a key investor.
The fundraising valued Mistral at 11.7 billion euros, and chief executive Arthur Mensch has said the company's revenue should top one billion euros this year.
mng/js/jhb

media

Instagram boss to testify at social media addiction trial

BY BENJAMIN LEGENDRE

  • YouTube and Meta -- the parent company of Instagram and Facebook -- are defendants in a blockbuster trial that could set a legal precedent regarding whether social media giants deliberately designed their platforms to be addictive to children.
  • Instagram chief Adam Mosseri is to be called to testify Wednesday in a Los Angeles courtroom by lawyers out to prove social media is dangerously addictive by design to young, vulnerable minds.
  • YouTube and Meta -- the parent company of Instagram and Facebook -- are defendants in a blockbuster trial that could set a legal precedent regarding whether social media giants deliberately designed their platforms to be addictive to children.
Instagram chief Adam Mosseri is to be called to testify Wednesday in a Los Angeles courtroom by lawyers out to prove social media is dangerously addictive by design to young, vulnerable minds.
YouTube and Meta -- the parent company of Instagram and Facebook -- are defendants in a blockbuster trial that could set a legal precedent regarding whether social media giants deliberately designed their platforms to be addictive to children.
Rival lawyers made opening remarks to jurors this week, with an attorney for YouTube insisting that the Google-owned video platform was neither intentionally addictive nor technically social media.
"It's not social media addiction when it's not social media and it's not addiction," YouTube lawyer Luis Li told the 12 jurors during his opening remarks.
The civil trial in California state court centers on allegations that a 20-year-old woman, identified as Kaley G.M., suffered severe mental harm after becoming addicted to social media as a child.
She started using YouTube at six and joined Instagram at 11, before moving on to Snapchat and TikTok two or three years later.
The plaintiff "is not addicted to YouTube. You can listen to her own words -- she said so, her doctor said so, her father said so," Li said, citing evidence he said would be detailed at trial.
Li's opening arguments followed remarks on Monday from lawyers for the plaintiffs and co-defendant Meta. 
On Monday, the plaintiffs' attorney Mark Lanier told the jury YouTube and Meta both engineer addiction in young people's brains to gain users and profits.
"This case is about two of the richest corporations in history who have engineered addiction in children's brains," Lanier said.
"They don't only build apps; they build traps." 
But Li told the six men and six women on the jury that he did not recognize the description of YouTube put forth by the other side and tried to draw a clear line between YouTube's widely popular video app and social media platforms like Instagram or TikTok.
YouTube is selling "the ability to watch something essentially for free on your computer, on your phone, on your iPad," Li insisted, comparing the service to Netflix or traditional TV.
Li said it was the quality of content that kept users coming back, citing internal company emails that he said showed executives rejecting a pursuit of internet virality in favor of educational and more socially useful content.

'Gateway drug'

Stanford University School of Medicine professor Anna Lembke, the first witness called by the plaintiffs, testified that she views social media, broadly speaking, as a drug.
The part of the brain that acts as a brake when it comes to having another hit is not typically developed before a person is 25 years old, Lembke, the author of the book "Dopamine Nation," told jurors.
"Which is why teenagers will often take risks that they shouldn't and not appreciate future consequences," Lembke testified.
"And typically, the gateway drug is the most easily accessible drug," she said, describing Kaley's first use of YouTube at the age of six.
The case is being treated as a bellwether proceeding whose outcome could set the tone for a wave of similar litigation across the United States.
Social media firms face hundreds of lawsuits accusing them of leading young users to become addicted to content and suffer from depression, eating disorders, psychiatric hospitalization, and even suicide.
Lawyers for the plaintiffs are borrowing strategies used in the 1990s and 2000s against the tobacco industry, which faced a similar onslaught of lawsuits arguing that companies knowingly sold a harmful product.
arp-gc/sla

technology

Google turns to century-long debt to build AI

  • While 100-year bonds are not new, it has been decades since US companies have resorted to them.
  • Google-parent Alphabet will issue bonds maturing in 100 years as it continues to invest massively in infrastructure for artificial intelligence, according to data published Tuesday by Bloomberg. 
  • While 100-year bonds are not new, it has been decades since US companies have resorted to them.
Google-parent Alphabet will issue bonds maturing in 100 years as it continues to invest massively in infrastructure for artificial intelligence, according to data published Tuesday by Bloomberg. 
The Silicon Valley internet giant reportedly aims to raise about $20 billion overall, a chunk of it by issuing bonds that mature in February of 2126, with lenders so keen for a piece of the AI action that some $100 billion orders were placed for the debt.
Alphabet did not respond to a request for comment.
Alphabet and AI race rivals including Amazon, Meta, Microsoft are investing staggering amounts in infrastructure to power the technology, banking on it paying off.
Market reaction, though, has been mixed with some investors worried spending has gone overboard.
Century-long bond issues by companies are a rarity, and especially for Alphabet which has ample online ad revenue available to pay for investments rather than resorting to debt.
But, the rush to lead in AI has changed the game, calling for unprecedented spending on data centers, energy generation and more.
Alphabet allocated $91 billion to spending on computing infrastructure last year and has told financial analysts it expects to spend from $175 billion to $185 billion on it this year.
Alphabet has ramped up longterm debt to handle the spending surge, issuing 50-year bonds late last year.
While 100-year bonds are not new, it has been decades since US companies have resorted to them.
Companies such as Disney, Coca-Cola, FedEx, Ford, and Motorola turned to such century-long debt during the 1990s.
tu-gc/arp

AI

Latam-GPT: a Latin American AI to combat US-centric bias

BY AXL HERNANDEZ

  • Unlike closed generative models like ChatGPT or Google's Gemini, Latam-GPT is an open model that can be used by programmers to customize parts of the software to suit their needs.
  • Move over ChatGPT. Chile on Tuesday launched Latam-GPT, an open-source artificial intelligence model for the region, designed to combat bias inherent in a US-centric industry.
  • Unlike closed generative models like ChatGPT or Google's Gemini, Latam-GPT is an open model that can be used by programmers to customize parts of the software to suit their needs.
Move over ChatGPT. Chile on Tuesday launched Latam-GPT, an open-source artificial intelligence model for the region, designed to combat bias inherent in a US-centric industry.
Developed by the Chilean National Center for Artificial Intelligence (CENIA), Latam-GPT uses millions of data points collected in Latin America to showcase the continent's cultural diversity. 
"Thanks to Latam-GPT, we’re positioning the region as an active and sovereign player in the economy of the future," President Gabriel Boric said of the initiative.
"We’re at the table -- we’re not on the menu," he added.
According to Chile's Science Minister Aldo Valle, the program was built to combat what he called prejudices and generalizations about people and countries from the region.
Latin America, he added, "cannot simply be a passive user or recipient of artificial intelligence systems. That could result in the loss of a significant part of our traditions."
Unlike closed generative models like ChatGPT or Google's Gemini, Latam-GPT is an open model that can be used by programmers to customize parts of the software to suit their needs.
Contributions to the project, and data for the model's training, were provided by Latin American universities, foundations, libraries, government entities and civil society organizations in countries including Argentina, Brazil, Chile, Colombia, Ecuador, Mexico, Peru and Uruguay.
"The models developed in other parts of the world do have data from Latin America but it represent a fairly small proportion," CENIA director Alvaro Soto noted. 
This low level of diverse input is sometimes reflected in the depictions of Latin Americans by major AI models. ChatGPT, for example, portrays a typical Chilean man as a person wearing a poncho with the Andes in the background.

Indigenous content

Major US tech companies dominate the global AI race, with low-cost Chinese models rapidly gaining ground and Europe lagging in third place.
Other regions of the world are also embracing the importance of developing public AI models that respect their cultural norms and safety standards.
In 2023, Singapore researchers released the open-source Southeast Asian Languages in One Network, or SEA-LION model, while in Kenya, the UlizaLLama LLM provides health services for Swahili-speaking expectant mothers.
Latam-GPT has been trained on more than eight terabytes of data, equivalent to millions of books. 
It was developed for a mere $550,000, sourced primarily from the Development Bank of Latin America (CAF) and CENIA's own resources.
A first version was developed on the Amazon Web Services cloud, but in future, Latam-GPT will be trained on a supercomputer at the University of Tarapaca in northern Chile.
For now, it is trained mainly in Spanish and Portuguese content, although its developers plan to incorporate material in Indigenous Latin American languages.

Slang and sayings

Latam-GPT will be available free of charge to companies and public institutions to develop applications more specific to Latin America, said Soto, the CENIA director. 
He cited potential applications for hospitals "with logistical problems or issues with the use of medical resources."
Its tiny budget means Latam-GPT has "no chance" of competing against the major AI models, Alejandro Barros, a professor in the Department of Industrial Engineering at the University of Chile, told AFP.
But it has already won over Chilean serial digital entrepreneur Roberto Musso, whose company Digevo plans to use Latam-GPT to develop customer service programs for airlines or retailers.
Musso said his clients were "very interested in having their users express themselves and receive responses in the local language." 
Latam-GPT, he said, provides the ability to recognize regional "slang, idioms, and even speech rate" and avoid biases that could arise in other AI models.
axl/cb/jgc/mlr/dw

Telegram

Moscow chokes Telegram as it pushes state-backed rival app

  • Moscow has been threatening various internet platforms with forced slowdowns or outright bans if they do not comply with Russian laws.
  • Russia's internet watchdog on Tuesday announced it was throttling the Telegram messenger platform for alleged legal violations, as Moscow tries to push its citizens into using a more tightly controlled domestic online service.
  • Moscow has been threatening various internet platforms with forced slowdowns or outright bans if they do not comply with Russian laws.
Russia's internet watchdog on Tuesday announced it was throttling the Telegram messenger platform for alleged legal violations, as Moscow tries to push its citizens into using a more tightly controlled domestic online service.
Moscow has been threatening various internet platforms with forced slowdowns or outright bans if they do not comply with Russian laws.
Those laws require data on Russian users to be stored inside the country, and for efforts to be made to stamp out their use for what Moscow calls "criminal and terrorist purposes".
Critics and rights campaigners say the restrictions are a transparent attempt by the Kremlin to ramp up control and surveillance over internet use in Russia, amid a sweeping crackdown on dissent during the Ukraine offensive.
Telegram's Russian-born founder Pavel Durov, who lives outside the country, posted on his Telegram channel that "Russia is restricting access to Telegram in an attempt to force its citizens to switch to a state-controlled app built for surveillance and political censorship".
The Roskomnadzor agency said in a statement cited by state media that it will "continue to introduce phased restrictions" on Telegram, which it said had not complied with the laws.
Media watchdog Reporters Without Borders (RSF) condemned what it called a continuous "strategy to strangle the circulation of information" and noted that Russia ranks 171st out of 180 in its World Press Freedom Index.
Amnesty International meanwhile branded the move "censorship and obstruction under the guise of protecting people's rights and interests".
Telegram is widely used across Russia, both as a messaging app and as a social media service.
Almost all major public figures, including government bodies and the Kremlin, post regular updates on the platform.
Some pro-war bloggers, who also use Telegram extensively, criticised the decision, saying it would hobble communications around the front line and in Russian-occupied territory.
"It's very unpleasant," said the Two Majors channel, one of Russia's most widely read military correspondents.
"People's positions will now mostly be conveyed to the outside world not by people, but by our masters of the foreign ministry," it added, lamenting the switch to Russian apps that nobody outside the country uses.

Max rival

Moscow is trying to push users onto a state-backed competitor, called Max, which can also handle payments and government services.
Pro-war correspondent Alexander Kots also said blocking Telegram would limit Russia's own "information operations", and recruitment of Ukrainians through the app to carry out sabotage attacks.
Both sides widely accuse each other of plotting behind-the-lines operations by recruiting sympathisers, or those in need of cash, over social media.
Before the war, Russia had previously tried to ban Telegram -- which is still run by Durov, who also possesses French and Emirati citizenships -- but ultimately failed in its attempts to block access and lifted the ban in 2020.
Russian users reported slow traffic and lagging downloads on Telegram throughout Tuesday before the official announcement.
Roskomnadzor has tried to choke other foreign services, including WhatsApp, owned by Facebook parent company Meta, and Google's YouTube.
Durov has previously clashed with Russian authorities.
He was forced out of the VK social media site he founded -- a Russian equivalent of Facebook -- under pressure from the authorities.
He went on to use the proceeds of the sale to launch Telegram in exile from the United Arab Emirates.
"Restricting citizens' freedom is never the right answer. Telegram stands for freedom of speech and privacy, no matter the pressure," he posted Tuesday.
Durov was detained in Paris in 2024 under a French investigation into Telegram's alleged complicity in criminal activity. France in July 2025 lifted travel restrictions on him but is keeping up its investigation.
bur/phz/cc/sbk/gv/sbk

Amazon

Europe's Ariane 6 to launch Amazon constellation satellites into orbit

BY OLGA NEDBAEVA

  • Rival Starlink, meanwhile, has nearly 9,400 satellites.
  • An enhanced version of Europe's Ariane 6 rocket will blast off Thursday to launch 32 satellites into orbit, forming part of the Amazon Leo network, which it hopes will rival Elon Musk's Starlink.
  • Rival Starlink, meanwhile, has nearly 9,400 satellites.
An enhanced version of Europe's Ariane 6 rocket will blast off Thursday to launch 32 satellites into orbit, forming part of the Amazon Leo network, which it hopes will rival Elon Musk's Starlink.
The launch, scheduled at 1645 GMT, will be a first for Amazon Leo, formerly known as Project Kuiper, from Europe's spaceport in Kourou, French Guiana, on the northeastern coast of South America. 
US firm Amazon, founded by billionaire Jeff Bezos, is the main commercial partner for the Ariane 6, despite the latter being touted as a symbol of European sovereignty in the sector.
"Over time a sovereign European launcher cannot be primarily dependent on foreign markets," warned Ludwig Moeller, director of the European Space Policy Institute (ESPI). 
Foreign partners "may negotiate priority handling backed by economic power or which may become unpredictable or inaccessible without notice, given the current geopolitical environment and trade wars," he told AFP.
But in the absence of European commercial customers -- many of whom work with Musk's SpaceX -- the Amazon partnership is crucial.
Four out of five anticipated launches took place in 2025 following Ariane's inaugural 2024 flight, unprecedented for a new launcher, according to ArianeGroup president Marc Sion.
Although Ariane 6 is eventually expected to carry out 10 launches per year, Pierre Lionnet, Eurospace research director, noted that at this stage this would not be possible without commercial customers like Amazon.
- Expansion - 
To take on Amazon Leo's 32 satellites, the Ariane 6 has been upgraded with four strap-on boosters, instead of the two used on the first five flights. 
It marks "our largest payload that we have launched to date," Martijn Van Delden, head of commercial development for Europe at Amazon Leo, told AFP. 
The upgrade is "impactful" by being more cost-effective and broadening fast internet networks to more customers, he said, noting it also "strengthens" the European space industry. 
With 175 satellites already in orbit, Amazon Leo aims to expand its constellation to 3,200. 
Rival Starlink, meanwhile, has nearly 9,400 satellites.
"We're looking to then increase the payload every time we have a new mission, especially as more powerful boosters come online on Ariane 6," Van Delden said. 
The rollout of its parent project, Amazon Leo, however, has faced challenges and delays. 
"Deploying 32 satellites is more complicated than deploying one -- you have to separate them one after another," Lionnet explained.
Long-term investment is expected to amount to billions of euros to the European space sector. 
"If things go well here, it will help build market confidence," said Philippe Clar, ArianeGroup's head of launchers.
neo/uh/nth/giv/rlp/ks/st

Telegram

Moscow chokes Telegram as it pushes state-backed rival app

  • Moscow has been threatening a host of internet platforms with forced slowdowns or outright bans if they do not comply with Russian laws.
  • Russia's internet watchdog on Tuesday announced it was throttling the Telegram messenger platform for alleged violations of Russian law, as Moscow tries to push Russians to use a more tightly controlled domestic online service.
  • Moscow has been threatening a host of internet platforms with forced slowdowns or outright bans if they do not comply with Russian laws.
Russia's internet watchdog on Tuesday announced it was throttling the Telegram messenger platform for alleged violations of Russian law, as Moscow tries to push Russians to use a more tightly controlled domestic online service.
Moscow has been threatening a host of internet platforms with forced slowdowns or outright bans if they do not comply with Russian laws.
Those laws require data on Russian users to be stored inside the country, and for efforts to be made to stamp out their use for what Moscow calls "criminal and terrorist purposes".
Critics and rights campaigners say those restrictions are a transparent attempt by the Kremlin to ramp up control and surveillance over internet use in Russia, amid a sweeping crackdown on dissent during the Ukraine offensive.
The Roskomnadzor agency said in a statement cited by state media that it will "continue to introduce phased restrictions" on Telegram, which it said had not complied with the laws.
Telegram is widely used across Russia, both as a messaging app and as a social media service.
Almost all major public figures, including government bodies and the Kremlin, post regular updates on the platform.
Moscow is trying to push users onto a state-backed competitor, called Max, which can also handle payments and government services.
Russia had previously tried to ban Telegram -- run by Pavel Durov, a Russian who also possesses French and Emirati nationalities --but ultimately failed in its attempts to block access and lifted the ban in 2020.
Russian users reported slow traffic and lagging downloads on Telegram throughout Tuesday before the official announcement.
Roskomnadzor has previously tried to choke other foreign services, including WhatsApp, owned by Facebook parent company Meta, and Google's YouTube.
Durov has previously clashed with Russian authorities.
He was forced out of the VK social media site he founded -- a Russian equivalent of Facebook -- under pressure from the authorities.
He went on to use the proceeds of the sale to launch Telegram in exile from the United Arab Emirates.
He was detained in Paris in 2024 under a French investigation into Telegram's alleged complicity in criminal activity. France in July 2025 lifted travel restrictions on Durov but is keeping up its investigation.
bur/rmb

AI

OpenAI starts testing ads in ChatGPT

  • "Ads do not influence the answers ChatGPT gives you, and we keep your conversations with ChatGPT private from advertisers," the company said.
  • OpenAI has begun placing ads in the basic versions of its ChatGPT chatbot, a bet that users will not mind the interruptions as the company seeks revenue as its costs soar.
  • "Ads do not influence the answers ChatGPT gives you, and we keep your conversations with ChatGPT private from advertisers," the company said.
OpenAI has begun placing ads in the basic versions of its ChatGPT chatbot, a bet that users will not mind the interruptions as the company seeks revenue as its costs soar.
"The test will be for logged-in adult users on the Free and Go subscription tiers" in the United States, OpenAI said Monday. The Go subscription costs $8 in the United States.
Only a small percentage of its nearly one billion users pay for its premium subscription services, which will remain ad-free.
"Ads do not influence the answers ChatGPT gives you, and we keep your conversations with ChatGPT private from advertisers," the company said.
Since ChatGPT's launch in 2022, OpenAI's valuation has soared to $500 billion in funding rounds -- higher than any other private company. Some analysts expect it could go public with a trillion-dollar valuation.
But the ChatGPT maker burns through cash at a furious rate, mostly on the powerful computing required to deliver its services.
Its chief executive Sam Altman had long expressed his dislike for advertising, citing concerns that it could create distrust about ChatGPT's content.
His about-face garnered a jab from its rival Anthropic over the weekend, which made its advertising debut at the Super Bowl championship with commercials saying its Claude chatbot would stay ad-free.
In one spot, a man asking an AI chatbot for advice on communicating with his mother receives earnest guidance before the conversation veers into a pitch for a fictional cougar-dating site called "Golden Encounters".
mng/jlo/js/jxb

media

Jury told that Meta, Google 'engineered addiction' at landmark US trial

BY GLENN CHAPMAN

  • Lawsuits, including some brought by school districts, accusing social media platforms of practices endangering young users are making their way through federal court in northern California and state courts across the country. 
  • Meta and Google-owned YouTube were accused Monday of pushing highly addictive apps on children as a landmark social media trial began in earnest in a California court. 
  • Lawsuits, including some brought by school districts, accusing social media platforms of practices endangering young users are making their way through federal court in northern California and state courts across the country. 
Meta and Google-owned YouTube were accused Monday of pushing highly addictive apps on children as a landmark social media trial began in earnest in a California court. 
The blockbuster trial in front of a Los Angeles jury could establish a legal precedent on whether the social media juggernauts deliberately designed their platforms to lead to addiction in children. 
The proceedings are expected to see Meta chief Mark Zuckerberg on the stand next week and Instagram boss Adam Mosseri in the courtroom as early as Wednesday. In addition to Instagram, Meta's platforms include Facebook and WhatsApp. 
"This case is about two of the richest corporations in history who have engineered addiction in children's brains," plaintiffs' attorney Mark Lanier told the jury in his opening statement. 
"This case is as easy as A-B-C," Lanier said as he stacked children's toy blocks bearing the letters. 
He contended the A was for addicting, the B for brains and the C for children. 
"They don't only build apps; they build traps," Lanier said, saying Meta and YouTube pursued "addiction by design," making his arguments using props like a toy Ferrari and a mini slot machine. 
Meta attorney Paul Schmidt countered in opening remarks to the jury that evidence will show problems with the plaintiff's family and real-world bullying took a toll on her self-esteem, body image and happiness rather than Instagram.
"If you took Instagram away and everything else was the same in Kaley's life, would her life be completely different, or would she still be struggling with the same things she is today?" Schmidt asked, pointing out an Instagram addiction is never mentioned in medical records included in the evidence.
The trial before Judge Carolyn Kuhl focuses on allegations that a 20-year-old woman identified as Kaley G.M. suffered severe mental harm because she became addicted to social media as a child. 
The case is being treated as a bellwether proceeding because its outcome could set the tone, and the level of payouts to successful plaintiffs, for a tidal wave of similar litigation across the United States. 
Social media firms are accused in hundreds of lawsuits of leading young users to become addicted to content and suffer from depression, eating disorders, psychiatric hospitalization and even suicide. 
Lawyers for the plaintiffs are borrowing strategies used in the 1990s and 2000s against the tobacco industry, which faced a similar onslaught of lawsuits arguing that companies knowingly sold a harmful product.
Lanier told the jurors that Kaley began watching YouTube at six years old because the company never told her mother "the goal was viewer addiction," or that toddlers as young as two were being targeted despite "critical" risk of addiction. 
"This is the first time that a social media company has ever had to face a jury for harming kids," Social Media Victims Law Center founder Matthew Bergman, whose team is involved in more than 1,000 such cases, told AFP.

'Strongly disagree'

Internet titans have argued that they are shielded by Section 230 of the US Communications Decency Act, which frees them from responsibility for what social media users post. 
However, this case argues that those firms are culpable for business models designed to hold people's attention and to promote content that can harm their mental health. 
The plaintiffs said they would call expert witnesses that will argue that young people's brains are not yet developed to withstand the powers of the algorithms being flung at them on Instagram and YouTube. 
The company pointed to recent efforts to provide more safeguards for young people, adding that "we're always working to do better." 
Jose Castaneda, a YouTube spokesperson, said "the allegations in these complaints are simply not true." 
Lawyers for YouTube are to present opening remarks to the jury on Tuesday.
Snapchat and TikTok were named as defendants in the suit, but struck settlement deals before the start of the trial. The terms were not disclosed. 
Lawsuits, including some brought by school districts, accusing social media platforms of practices endangering young users are making their way through federal court in northern California and state courts across the country. 
A separate lawsuit accusing Meta of putting profit over the wellbeing of young users was also getting under way in New Mexico on Monday.
gc-arp/jgc/mlm

AI

AI chatbots give bad health advice, research finds

  • People using the AI chatbots were only able to identify their health problem around a third of the time, while only around 45 percent figured out the right course of action.
  • Next time you're considering consulting Dr ChatGPT, perhaps think again.
  • People using the AI chatbots were only able to identify their health problem around a third of the time, while only around 45 percent figured out the right course of action.
Next time you're considering consulting Dr ChatGPT, perhaps think again.
Despite now being able to ace most medical licensing exams, artificial intelligence chatbots do not give humans better health advice than they can find using more traditional methods, according to a study published on Monday.
"Despite all the hype, AI just isn't ready to take on the role of the physician," study co-author Rebecca Payne from Oxford University said.
"Patients need to be aware that asking a large language model about their symptoms can be dangerous, giving wrong diagnoses and failing to recognise when urgent help is needed," she added in a statement.
The British-led team of researchers wanted to find out how successful humans are when they use chatbots to identify their health problems and whether they require seeing a doctor or going to hospital.
The team presented nearly 1,300 UK-based participants with 10 different scenarios, such as a headache after a night out drinking, a new mother feeling exhausted or what having gallstones feels like.
Then the researchers randomly assigned the participants one of three chatbots: OpenAI's GPT-4o, Meta's Llama 3 or Command R+. There was also a control group that used internet search engines. 
People using the AI chatbots were only able to identify their health problem around a third of the time, while only around 45 percent figured out the right course of action.
This was no better than the control group, according to the study, published in the Nature Medicine journal.

Communication breakdown

The researchers pointed out the disparity between these disappointing results and how AI chatbots score extremely highly on medical benchmarks and exams, blaming the gap on a communication breakdown.
Unlike the simulated patient interactions often used to test AI, the real humans often did not give the chatbots all the relevant information. 
And sometimes the humans struggled to interpret the options offered by the chatbot, or misunderstood or simply ignored its advice.
One out of every six US adults ask AI chatbots about health information at least once a month, the researchers said, with that number expected to increase as more people adopt the new technology.
"This is a very important study as it highlights the real medical risks posed to the public by chatbots," David Shaw, a bioethicist at Maastricht University in the Netherlands who was not involved in the research, told AFP.
He advised people to only trust medical information from reliable sources, such as the UK's National Health Service.
dl/sbk

internet

YouTube star MrBeast buys youth-focused banking app

  • Though it offers credit and debit cards, and interest-bearing accounts, Step is technically not a bank, but rather a financial services platform backed by Evolve Bank & Trust.
  • MrBeast, the world's most popular YouTuber, has purchased a Gen Z-focused banking app, he said Monday, marking another addition to his sprawling business interests.
  • Though it offers credit and debit cards, and interest-bearing accounts, Step is technically not a bank, but rather a financial services platform backed by Evolve Bank & Trust.
MrBeast, the world's most popular YouTuber, has purchased a Gen Z-focused banking app, he said Monday, marking another addition to his sprawling business interests.
The online star, whose real name is Jimmy Donaldson, announced on social media that his company Beast Industries acquired Step, a financial services platform.
"Nobody taught me about investing, building credit, or managing money when I was growing up," the 27-year-old said. "I want to give millions of young people the financial foundation I never had."
Though it offers credit and debit cards, and interest-bearing accounts, Step is technically not a bank, but rather a financial services platform backed by Evolve Bank & Trust.
It was not revealed how much the purchase cost. Step and Beast industries did not immediately respond to AFP's request for comment.
MrBeast boasts more than 450 million YouTube subscribers -- the most in the world -- with his channel specializing in absurd contests and globe-trotting philanthropic ventures.
Videos can cost millions of dollars to produce, with a team of around 300 people.
He has spun that success into a television game show on Amazon Prime Video, as well as a pop-up amusement park in Saudi Arabia dubbed Beast Land.
His snack brand, Feastables, meanwhile rakes in hundreds of millions in sales, rivalling the revenue from the YouTube videos that first made him famous.
"We're excited about how this acquisition is going to amplify our platform and bring more groundbreaking products to Step customers," Step CEO and founder CJ MacDonald said in a statement.
The contestants on MrBeast's videos could perhaps benefit from Step's goal of teaching financial literacy.
Recent posts include: "Survive 20 Days Chained To Your Ex, Win $250,000" and "30 Celebrities Fight For $1,000,000!"
tu/nro/mlm

Discord

Discord adopts facial recognition in child safety crackdown

  • Gaming platform Roblox in January began requiring facial age verification globally for all users to access chat features, after facing multiple lawsuits alleging the platform enabled predatory behavior and child exploitation.
  • Messaging platform Discord announced Monday it will implement enhanced safety features for teenage users globally, including facial recognition, joining a wave of social media companies rolling out age verification systems.
  • Gaming platform Roblox in January began requiring facial age verification globally for all users to access chat features, after facing multiple lawsuits alleging the platform enabled predatory behavior and child exploitation.
Messaging platform Discord announced Monday it will implement enhanced safety features for teenage users globally, including facial recognition, joining a wave of social media companies rolling out age verification systems.
The rollout, beginning in early March, will make teen-appropriate settings the default for all users, with adults needing to verify their age to loosen protections including content filters and bans on direct messaging, the company said.
The San Francisco-based platform, popular among gamers, will use facial age estimation technology and identity verification through vendor partners to determine users' ages.
Tracking software running in the background will also help determine the age of users without always requiring direct verification.
"Nowhere is our safety work more important than when it comes to teen users," said Savannah Badalich, Discord's head of product policy.
Discord insisted the measures came with privacy protections, saying video selfies for age estimation never leave users' devices and that submitted identity documents are deleted quickly.
The platform said it successfully tested the measures in Britain and Australia last year before expanding worldwide.
The move follows similar actions by rivals facing intense scrutiny over child safety and follows an Australian ban on under-16s using social media that is being duplicated in other countries.
Resorting to facial recognition and other technologies addresses the reality that self-reported age has proven unreliable, with minors routinely lying about their birthdates to circumvent platform safety measures.
Gaming platform Roblox in January began requiring facial age verification globally for all users to access chat features, after facing multiple lawsuits alleging the platform enabled predatory behavior and child exploitation.
Meta, which owns Instagram and Facebook, has deployed AI-powered methods to determine age and introduced "Teen Accounts" with automatic restrictions for users under 18.
Mark Zuckerberg's company removed over 550,000 underage accounts in Australia alone in December ahead of that country's under-16 social media ban.
TikTok has implemented 60-minute daily screen time limits for users under 18 and notification cutoffs based on age groups.
The industry-wide shift comes as half of US states have enacted or introduced legislation involving age-related social media regulation, though courts have blocked many of the restrictions on free speech grounds.
The changes come the same day as a trial in California on social media addiction for children begins in Los Angeles, with plaintiffs alleging Meta's and YouTube's platforms were designed to be addictive to minors.
arp/des

telecommunication

Brazil seeks to restore block of Rumble video app

  • It was one of the reasons Trump cited when he imposed high tariffs on Brazilian products in 2025, and imposed sanctions on top judicial authorities.
  • Brazilian authorities were on Monday working to cut off access to the Rumble video app after it bypassed a ban imposed as part of the country's battle against disinformation.
  • It was one of the reasons Trump cited when he imposed high tariffs on Brazilian products in 2025, and imposed sanctions on top judicial authorities.
Brazilian authorities were on Monday working to cut off access to the Rumble video app after it bypassed a ban imposed as part of the country's battle against disinformation.
US-based Rumble, popular with conservative and far-right voices, was banned by Brazil's supreme court last year after it refused to block a Brazilian user living in the US who was accused of spreading disinformation.
Brazil's telecoms regulator Anatel said in a statement sent to AFP that it was working to "reinstate the block" after Rumble managed to hide its internet addresses by routing its service through another company's network.
In 2024, Elon Musk's X social network used a similar method to briefly evade a 40-day ban imposed by the Supreme Court over its failure to comply with a series of court orders against online disinformation.
Anatel said Rumble should again be off the air in a few days.
Brazil's crackdown on online disinformation angered US President Donald Trump, who accused the country of attacking American social media companies.
It was one of the reasons Trump cited when he imposed high tariffs on Brazilian products in 2025, and imposed sanctions on top judicial authorities.
Many of these measures have since been repealed amid a warming of diplomatic ties after Trump met with his Brazilian counterpart Luiz Inacio Lula da Silva in October.
jss/app/fb/dw

disinformation

'Flood' of disinformation ahead of Bangladesh election

BY EYAMIN SAJID WITH SUMIT DUBEY IN NEW DELHI

  • But while analysts say much of the disinformation originates from India, there is no evidence that the large-scale media posts were organised by the government.
  • Voters in Bangladesh elect a new government on February 12, but analysts warn their choice is threatened by a coordinated surge of disinformation, much of which originates from neighbouring India.
  • But while analysts say much of the disinformation originates from India, there is no evidence that the large-scale media posts were organised by the government.
Voters in Bangladesh elect a new government on February 12, but analysts warn their choice is threatened by a coordinated surge of disinformation, much of which originates from neighbouring India.
The Muslim-majority nation of around 170 million people is preparing for its first election since a 2024 student-led uprising toppled Sheikh Hasina -- who fled to neighbouring India, where she has been hosted since by the Hindu-nationalist government.
Authorities say the scale of online manipulation -- including sophisticated AI-generated images --  has become so severe that a special unit has been created to curb false content.
Interim leader and Nobel Peace Prize winner Muhammad Yunus said in January that there had been a "flood of misinformation surrounding the elections" when he called UN rights chief Volker Turk seeking help. 
"It is coming from both foreign media and local sources," he said.
Much of that centres around claims of attacks against Bangladesh's minorities -- around 10 percent of Bangladesh's population is non-Muslim, most of them Hindu.
That has seen a mass posting of claims online that Hindus are under attack, using the hashtag "Hindu genocide".
According to police figures released in January, out of 645 incidents involving members of minority groups in 2025 -- only 12 percent were classified as having a sectarian motive.

'Coordinated Indian disinformation'

The US-based Center for the Study of Organized Hate said it had tracked more than 700,000 posts -- generated by more than 170,000 accounts on X, that made claims of a "Hindu genocide" between August 2024 and January 2026.
"We have tracked coordinated Indian disinformation online, falsely alleging large-scale violence against Hindus in Bangladesh," said Raqib Naik, head of the think tank.
"More than 90 percent of this content originated from India, with the remainder linked to associated Hindu nationalist networks in the UK, US, and Canada," he told AFP.
Examples debunked by AFP Fact Check, some of them shared tens of thousands of times, include an AI-created video of a woman who had lost her arm, appealing not to vote for the Bangladesh Nationalist Party (BNP), seen by many as a frontrunner.
In another computer-generated video, a Hindu woman alleges that people who follow the same religion have been told to vote for Jamaat-e-Islami, the key Islamist party, or they will be exiled to India.
Of the hundreds of AI-generated videos documented by AFP Fact Check teams on social media platforms –- YouTube, Facebook, TikTok and Instagram -- few are marked with an AI disclaimer.
The surge has also come after years of repression under Hasina, when opposition was crushed and outspoken voices silenced.
"We are noticing a huge amount of fake information compared to other times," said Miraj Ahmed Chowdhury, head of the Dhaka-based research organisation Digitally Right, saying free AI tools made creating sophisticated fakes easier. 
In another AI-generated video, Bangladeshis appear to praise Hasina -- now a fugitive who was sentenced to death in absentia for crimes against humanity.
In India, social media outrage by Hindu fundamentalists about the lone Bangladeshi cricket player in India's domestic IPL league resulted in his club cancelling his contract -- a furore that escalated to Bangladesh's national team pulling out of this month's T20 World Cup in India.
But while analysts say much of the disinformation originates from India, there is no evidence that the large-scale media posts were organised by the government.
New Delhi's foreign ministry say they have recorded a "disturbing pattern of recurring attacks on minorities" by "extremists in Bangladesh", but also emphasise they have "consistently reiterated our position in favour of free, fair, inclusive and credible elections".

'Big threat'

Bangladesh Election Commission spokesman Md. Ruhul Amin Mallik said they were working with Facebook's parent company, Meta, and set up a unit to monitor social media posts -- but coping with the sheer volume online is a never-ending task. 
"If our team detects any content as harmful and misleading, we instantly announce it as fake information," Mallik said.
Election expert Jasmine Tuli, a former election commission official, said that AI-generated images carried an extra risk for Bangladesh.
More than 80 percent of urban households have at least one smartphone, and nearly 70 percent of rural areas, according to government statistics -- but many people are still relatively new to the technology.
"It is a big threat for a country like Bangladesh, since people don't have much awareness to check the information," Tuli said. 
"Due to AI-generated fake visuals, voters get misguided in their decision."
burs-pjm/ane

crime

French police arrest six over crypto-linked magistrate kidnapping

BY MANON BILLING AND SABINE COLPART

  • Four men and one woman were detained, three overnight and two on Sunday morning, Lyon prosecutor Thierry Dran told AFP. He later confirmed a minor had been arrested on Sunday afternoon. 
  • French authorities have arrested six suspects, including a minor, after a magistrate and her mother were held captive last week for around 30 hours in a cryptocurrency ransom plot, prosecutors said on Sunday.
  • Four men and one woman were detained, three overnight and two on Sunday morning, Lyon prosecutor Thierry Dran told AFP. He later confirmed a minor had been arrested on Sunday afternoon. 
French authorities have arrested six suspects, including a minor, after a magistrate and her mother were held captive last week for around 30 hours in a cryptocurrency ransom plot, prosecutors said on Sunday.
Four men and one woman were detained, three overnight and two on Sunday morning, Lyon prosecutor Thierry Dran told AFP. He later confirmed a minor had been arrested on Sunday afternoon. 
The individuals were taken into custody following the discovery of the 35-year-old magistrate and her 67-year-old mother on Friday morning, found injured in a garage in the southeastern Drome region.
Two of those arrested overnight were detained as they attempted to take a bus to Spain, according to a source close to the case speaking on condition of anonymity.
Authorities continue to actively search for further suspects, a second source close to the case said, adding that the woman in custody is the partner of one of the four male suspects.
During a press conference Friday after the pair's escape, prosecutor Dran said the magistrate's partner -- who was not home when the two victims were abducted overnight Wednesday to Thursday -- has a leading position in a cryptocurrency start-up.
A massive police search involving 160 officers was launched after the magistrate's partner had received a message and a photo of her from the kidnappers demanding a ransom to be paid in cryptocurrency. 
The captors threatened to mutilate the victims if the transfer was not made quickly, Dran told reporters, declining to specify the amount demanded.
But the two women managed to free themselves and call for help without any ransom being paid, by banging on the garage door in Bourg-les-Valence. 
"Alerted by the noise, a neighbour intervened. He was able to open the door and allow our two victims to escape," Dran said.

Crypto-linked kidnappings

French authorities have been dealing with a string of kidnappings and extortion attempts targeting the families of wealthy individuals dealing in cryptocurrencies.
In January 2025, kidnappers seized French crypto boss David Balland and his partner. Balland co-founded a crypto firm called Ledger, valued at the time at more than $1 billion.
Balland's kidnappers cut off his finger and demanded a hefty ransom. He was freed the next day, and his girlfriend was found tied up in the boot of a car outside Paris.
In May, the father of a man who ran a Malta-based cryptocurrency company was kidnapped by four hooded men in Paris. 
The victim, whose finger was also severed by the kidnappers and for whom a ransom of several million euros was demanded, was released 58 hours later in a raid by the security forces.
sc-mlb/ekf/rh/gv

crime

French police arrest five over crypto-linked magistrate kidnapping

  • The arrests of four men and one woman followed the discovery of the 35-year-old magistrate and her 67-year-old mother on Friday, found injured in a garage in the southeastern Drome department, the Lyon public prosecutor's office said.
  • French authorities have arrested five suspects after a magistrate and her mother were held captive last week for around 30 hours in a cryptocurrency ransom plot, prosecutors told AFP on Sunday.
  • The arrests of four men and one woman followed the discovery of the 35-year-old magistrate and her 67-year-old mother on Friday, found injured in a garage in the southeastern Drome department, the Lyon public prosecutor's office said.
French authorities have arrested five suspects after a magistrate and her mother were held captive last week for around 30 hours in a cryptocurrency ransom plot, prosecutors told AFP on Sunday.
The arrests of four men and one woman followed the discovery of the 35-year-old magistrate and her 67-year-old mother on Friday, found injured in a garage in the southeastern Drome department, the Lyon public prosecutor's office said.
During a press conference held later on Friday, Lyon prosecutor Thierry Dran said the magistrate's partner -- who was not home when the pair were abducted overnight Wednesday to Thursday -- has a leading position in a cryptocurrency start-up.
A massive police search involving 160 officers was launched after he had received a message and a photo of his partner from the kidnappers demanding a ransom to be paid in cryptocurrency. 
The captors threatened to mutilate the victims if the transfer was not made quickly, Dran told reporters, declining to specify the amount demanded.
But the two women managed to free themselves and raise the alarm. They were rescued Friday morning in Bourg-les-Valence without any ransom being paid, according to the prosecutor.
French authorities have been dealing with a string of kidnappings and extortion attempts targeting the families of wealthy individuals dealing in cryptocurrencies.
In January 2025, kidnappers seized French crypto boss David Balland and his partner. Balland co-founded a crypto firm called Ledger, valued at the time at more than $1 billion.
Balland's kidnappers cut off his finger and demanded a hefty ransom. He was freed the next day, and his girlfriend was found tied up in the boot of a car outside Paris.
In May, the father of a man who ran a Malta-based cryptocurrency company was kidnapped by four hooded men in Paris. 
The victim, whose finger was also severed by the kidnappers and for whom a ransom of several million euros was demanded, was released 58 hours later in a raid by the security forces.
sc-mlb/ekf/rmb

media

Opinions of Zuckerberg hang over social media addiction trial jury selection

BY BENJAMIN LEGENDRE

  • - 'Start fairly' - Jury selection was dominated by recurring references to Zuckerberg, the head of Meta and co-founder of Facebook who reached global fame after the Hollywood film "The Social Network."
  • A jury has been confirmed in a landmark social media addiction trial in the US state of California, a process dominated by references to tech giant Meta's divisive founder Mark Zuckerberg.
  • - 'Start fairly' - Jury selection was dominated by recurring references to Zuckerberg, the head of Meta and co-founder of Facebook who reached global fame after the Hollywood film "The Social Network."
A jury has been confirmed in a landmark social media addiction trial in the US state of California, a process dominated by references to tech giant Meta's divisive founder Mark Zuckerberg.
Meta's lawyers fought for six days in court to remove jurors who they deemed overly hostile to Facebook and Instagram, two of the social media platforms involved in the case.
The plaintiff's lawyers sought to dismiss people, mostly men, who believed that young internet users' mental health issues are more attributable to parental failures rather than tech platform designers.
With the jury of 12 members and six alternates approved on Friday, arguments in the case are now scheduled to begin Monday in Los Angeles Superior Court.
The case is being called a bellwether proceeding because its outcome could set the tone for a tidal wave of similar litigation across the United States.
Defendants at the trial are Alphabet and Meta, the tech titans behind YouTube and Instagram. TikTok and Snapchat were also accused, but have since settled for an undisclosed amount.
The trial focuses on allegations that a 20-year-old woman identified by the initials K.G.M. suffered severe mental harm because she became addicted to social media as a child.
She accuses Meta and YouTube of knowingly designing addictive apps, to the detriment of her mental health. 

'Start fairly'

Jury selection was dominated by recurring references to Zuckerberg, the head of Meta and co-founder of Facebook who reached global fame after the Hollywood film "The Social Network."
"I feel impartial toward the plaintiff, but based on things Mark Zuckerberg has done objectively -- I have strong feelings about -- and I think the defendant would start further behind," said one young woman.
Many potential jurors criticized Facebook's early days -- it was designed as a platform for college students to rate women's looks -- and cited the Cambridge Analytica privacy breach of 2018.
They also said it would be difficult for them to accept the billionaire's testimony -- expected in the next two weeks -- without prejudice.
Meta's lawyer, Phyllis Jones, raised frequent objections to such jurors.
She said it was "very important that both sides start fairly, with no disadvantage, that you look at the evidence fairly and decide."
Others were dismissed for the opposite reason.
"I like this guy," said one rare Zuckerberg fan. "I regret not owning Meta shares." 
He was dismissed by the plaintiff's lawyer, Mark Lanier.
Others to be removed included a man who expressed his anger against psychiatrists, and several people whose loved ones suffered from social media addiction or harassment.

Seeking distance

Alphabet's lawyers were keen to ensure that their platform YouTube was not lumped in with Meta.
"Does everybody understand that YouTube and Meta are very different companies? Does everyone understand that (Zuckerberg) doesn't run YouTube?" asked Luis Li, a lawyer for Google's video platform.
One man said he saw the potential for YouTube to seek to trigger "immediate dopamine" rushes among users through its "Shorts" feature.
He said his niece spends too much time on TikTok, which popularized a platform that provides endless scrolling of ultra-short-format videos.
The case will focus not on content, on which front platforms are largely protected by US law, but on the design of algorithms and personalization features.
The plaintiffs allege that the platforms are negligent and purposely designed to be harmful, echoing a strategy successfully used against the tobacco industry.
Meta and YouTube strongly deny the allegations, and also unsuccessfully argued on Friday for the judge to declare statements comparing their platforms to tobacco and other addictive products to be illegitimate.
The debate on the platform's level of responsibility for their effect on users was already underway, even at this early stage of the trial.
Alphabet's lawyer Li asked the panel if people spend too much time on phones, with the majority nodding in agreement.
"As a society, is it a problem?" he asked, with most hands again going up.
He then asked if this is "because of YouTube?" prompting hesitation from the jurors.
bl/aha/mlm

Global Edition

Crypto firm accidentally sends $40 bn in bitcoin to users

  • Bithumb said it accidentally sent 620,000 bitcoins, currently worth more than $40 billion, and blocked trading and withdrawals for the 695 affected users within 35 minutes after the error occurred on Friday.
  • A South Korean cryptocurrency exchange apologised on Saturday after mistakenly transferring more than $40 billion worth of bitcoin to users, which briefly prompted a selloff on the platform.
  • Bithumb said it accidentally sent 620,000 bitcoins, currently worth more than $40 billion, and blocked trading and withdrawals for the 695 affected users within 35 minutes after the error occurred on Friday.
A South Korean cryptocurrency exchange apologised on Saturday after mistakenly transferring more than $40 billion worth of bitcoin to users, which briefly prompted a selloff on the platform.
Bithumb said it accidentally sent 620,000 bitcoins, currently worth more than $40 billion, and blocked trading and withdrawals for the 695 affected users within 35 minutes after the error occurred on Friday.
According to local reports, Bithumb was meant to send about 2,000 won ($1.37) to each customer as part of a promotion, but mistakenly transferred roughly 2,000 bitcoins per user.
"We sincerely apologise for the inconvenience caused to our customers due to the confusion that occurred during the distribution process of this (promotional) event," Bithumb said in a statement.
The platform said it had recovered 99.7 percent of the mistakenly sent bitcoins, and that it would use its own assets to fully cover the amount that was lost in the incident.
It admitted the error briefly caused "sharp volatility" in bitcoin prices on the platform as some recipients sold the tokens, adding that it brought the situation under control within five minutes.
Its charts showed the token's prices briefly went down 17 percent to 81.1 million won on the platform late Friday.
In a separate statement released later on Saturday, Bithumb said some trades were executed at unfavourable prices for users due to a price drop during the incident Friday, including "panic selling".
The platform said it would compensate affected customers, covering the full price difference as well as a 10-percent bonus.
It estimated losses at about 1 billion won.
The platform earlier stressed that the incident was "unrelated to external hacking or security breaches".
Bitcoin, the world's biggest cryptocurrency, sank this week, wiping out gains sparked by US President Donald Trump's presidential election victory in November 2024.
cdl/ami

children

EU tells TikTok to change 'addictive' design

BY RAZIYE AKKOC

  • EU tech chief Henna Virkkunen told reporters that "TikTok has to take actions, they have to change the design of their service in Europe to protect our minors and their wellbeing".
  • The EU said Friday that TikTok needs to change its "addictive design" or risk heavy fines under the bloc's digital content rules, drawing a sharp pushback from the Chinese-owned platform.
  • EU tech chief Henna Virkkunen told reporters that "TikTok has to take actions, they have to change the design of their service in Europe to protect our minors and their wellbeing".
The EU said Friday that TikTok needs to change its "addictive design" or risk heavy fines under the bloc's digital content rules, drawing a sharp pushback from the Chinese-owned platform.
In preliminary conclusions of a probe opened two years ago, the European Commission said it found TikTok was not taking effective steps to address the app's negative impacts, especially on minors and vulnerable adults.
"TikTok's addictive design is in breach of the Digital Services Act," said commission spokesperson Thomas Regnier, citing concerns with features such as infinite scroll, autoplay, push notifications, and a highly personalised recommender system.
"These features lead to the compulsive use of the app, especially for our kids, and this poses major risks to their mental health and wellbeing," Regnier said, adding: "The measures that TikTok has in place are simply not enough."
TikTok rejected the commission's findings, with a spokesperson saying it presented "a categorically false and entirely meritless depiction of our platform".
"We will take whatever steps are necessary to challenge these findings through every means available to us," they added in a statement.
Allies of US President Donald Trump in the US Congress said the European Commission's "punitive actions" were a pretext for curbing political speech and pressuring companies.
The DSA is part of a bolstered legal armoury adopted by the EU in recent years to curb Big Tech's excesses, and officials had until now said TikTok was cooperating with the bloc's digital regulators.
TikTok will now have access to the EU's findings in order to defend itself against the claims.
EU tech chief Henna Virkkunen told reporters that "TikTok has to take actions, they have to change the design of their service in Europe to protect our minors and their wellbeing".
The commission gave examples of what the platform could alter, such as: 
-- the platform's "infinite scroll" offering users an uninterrupted feed
-- implementing effective "screen time breaks", including during the night
-- adapting its recommender system, the algorithms used by platforms to feed users more personalised content.

'Compulsive use' of TikTok

The February 2024 investigation was the first opened into TikTok under the DSA, the bloc's powerful content moderation law that has faced the wrath of the US administration under President Donald Trump.
In presenting the probe findings, Regnier cited what he called "extremely alarming" statistics on the app's use in the EU.
TikTok was "by far" the most-used platform after midnight by children between 13 and 18, he said, with seven percent of children aged 12 to 15 spending four to five hours daily on it. 
Brussels accused TikTok of disregarding "important indicators of compulsive use of the app" such as the time minors spent on the platform at night.
It also said TikTok had not implemented effective measures to mitigate risks, taking particular aim at screen time management and parental control tools.
Its time management tools were "easy to dismiss" including for young users, the commission found, while parental controls required "additional time and skills from parents to introduce" them.

'Safe by design'

The findings come as several European countries move to curb access to social media for younger teenagers, with officials weighing whether it is time to follow suit at EU level.
Briefing reporters Friday, Virkkunen said her priority was to make platforms safe for all users, children included.
"Social media should be so safe by design that we shouldn't have that kind of very high age restriction," she said.
If the regulator's views on TikTok are confirmed, the commission can impose a fine of up to six percent of the company's total worldwide annual turnover.
The EU began a separate probe into TikTok in December 2024 on alleged foreign interference during the Romanian presidential elections.
EU spokesman Regnier said earlier this week that TikTok was "extremely cooperative" during that investigation and was taking measures to address the commission's concerns.
He added that while the probe remained open, regulators could monitor how TikTok behaves during other elections.
raz-ec/js/phz