विज्ञान

By Olivia Le Poidevin GENEVA (Reuters) -Leading AI assistants misrepresent news content in nearly half their responses, according to new research published on Wednesday by the European Broadcasting Union (EBU) and the BBC. The international research studied 3,000 responses to questions about the news from leading artificial intelligence assistants – software applications that use AI to understand natural language commands to complete tasks for a user. It assessed AI assistants in 14 languages for accuracy, sourcing and ability to distinguish opinion versus fact, including ChatGPT, Copilot, Gemini and Perplexity. Overall, 45% of the AI responses studied contained at least one significant issue, with 81% having some form of problem, the research showed. Reuters has made contact with the companies to seek their comment on the findings. Gemini, Google's AI assistant, has stated previously on its website that it welcomes feedback so that it can continue to improve the platform and make it more helpful to users. OpenAI and Microsoft have previously said hallucinations – when an AI model generates incorrect or misleading information, often due to factors such as insufficient data – are an issue that they are seeking to resolve. Perplexity says on its website that one of its "Deep Research" modes has 93.9% accuracy in terms of factuality. SOURCING ERRORS A third of AI assistants' responses showed serious sourcing errors such as missing, misleading or incorrect attribution, according to the study. Some 72% of responses by Gemini, Google's AI assistant, had significant sourcing issues, compared to below 25% for all other assistants, it said. Issues of accuracy were found in 20% of responses from all AI assistants studied, including outdated information, it said. Examples cited by the study included Gemini incorrectly stating changes to a law on disposable vapes and ChatGPT reporting Pope Francis as the current Pope several months after his death. Twenty-two public-service media organisations from 18 countries including France, Germany, Spain, Ukraine, Britain and the United States took part in the study. With AI assistants increasingly replacing traditional search engines for news, public trust could be undermined, the EBU said. "When people don’t know what to trust, they end up trusting nothing at all, and that can deter democratic participation," EBU Media Director Jean Philip De Tender said in a statement. Some 7% of all online news consumers and 15% of those aged under 25 use AI assistants to get their news, according to the Reuters Institute’s Digital News Report 2025. The new report urged AI companies to be held accountable and to improve how their AI assistants respond to news-related queries. (Reporting by Olivia Le Poidevin, Editing by Timothy Heritage)

(The article has been published through a syndicated feed. Except for the headline, the content has been published verbatim. Liability lies with original publisher.)

(Reuters) -Germany's Northern Data said on Tuesday that it had withdrawn its annual forecast, as the AI cloud company was evaluating potential strategic transactions and graphics processing unit's market pricing dynamics. Northern Data said in a statement that the forecast withdrawal is partially offset by improved utilization of its GPU capacity driven by customer traction after the second-quarter technology upgrade. The company said that its Taiga cloud business initiated an upgrade of its infrastructure in March to enable access to its GPU estate, which led to improving the company's ability to serve existing customers and added a new and diversified customer base. Currently, more than 15,000 GPUs of Northern Data's 22,000 H100 and H200 GPU estate have been allocated to customers, Northern Data added. Video platform Rumble, which hosts U.S. President Donald Trump's Truth Social, in August made an offer to acquire Northern Data, giving Rumble control of the German company's Taiga business and its large-scale data center arm, Ardent. Reuters calculated the potential total deal value at approximately $1.17 billion. (Reporting by Rishabh Jaiswal in Bengaluru; Editing by Anil D'Silva and Alan Barona)

(The article has been published through a syndicated feed. Except for the headline, the content has been published verbatim. Liability lies with original publisher.)

(Reuters) -Germany's Northern Data said on Tuesday that it had withdrawn its annual forecast, as the AI cloud company was evaluating potential strategic transactions and graphics processing unit's market pricing dynamics. Northern Data said in a statement that the forecast withdrawal is partially offset by improved utilization of its GPU capacity driven by customer traction after the second-quarter technology upgrade. The company said that its Taiga cloud business initiated an upgrade of its infrastructure in March to enable access to its GPU estate, which led to improving the company's ability to serve existing customers and added a new and diversified customer base. Currently, more than 15,000 GPUs of Northern Data's 22,000 H100 and H200 GPU estate have been allocated to customers, Northern Data added. Video platform Rumble, which hosts U.S. President Donald Trump's Truth Social, in August made an offer to acquire Northern Data, giving Rumble control of the German company's Taiga business and its large-scale data center arm, Ardent. Reuters calculated the potential total deal value at approximately $1.17 billion. (Reporting by Rishabh Jaiswal in Bengaluru; Editing by Anil D'Silva and Alan Barona)

(The article has been published through a syndicated feed. Except for the headline, the content has been published verbatim. Liability lies with original publisher.)

(Reuters) -AI startup Anthropic is in talks with Google to secure additional computing power valued in the high tens of billions of dollars, Bloomberg News reported on Tuesday, citing people familiar with the matter. The deal, which has not been finalized yet, includes Alphabet's Google providing cloud computing services to Anthropic, according to the report. Alphabet's stock gained 2.3% in trading after the bell. Google declined to comment, while Anthropic did not immediately respond to a request for comment. Anthropic, which counts Google and Amazon.com as its major investors, is the creator of the Claude chatbot and competes with OpenAI's ChatGPT for AI dominance. Anthropic is projecting to more than double and potentially nearly triple its annualized revenue run rate next year, fueled by the rapid adoption of its enterprise products, Reuters had exclusively reported last week. The company is on track to meet an internal goal of $9 billion in annual revenue run rate – a calculation of annual revenue extrapolated from the current sales pace – by the end of 2025. (Reporting by Juby Babu in Mexico City; Editing by Anil D'Silva)

(The article has been published through a syndicated feed. Except for the headline, the content has been published verbatim. Liability lies with original publisher.)

(Reuters) -Crypto heavyweight Coinbase said on Tuesday it has bought investment platform Echo in a nearly $375 million cash-and-stock deal, aiming to bring fundraising tools to its platform. Dealmaking within the digital assets industry has picked up pace this year as a crypto-friendly Trump administration encourages companies to expand their business in the U.S. Last week, cryptocurrency exchange Kraken unveiled a $100 million deal for futures exchange Small Exchange, paving the way to launch a fully U.S.-based derivatives suite. Echo's platform makes raising capital and investing more accessible to the crypto community through private and public token sales. "We want to create more accessible, efficient, and transparent capital markets," Coinbase said in a blog post. While Coinbase will start with crypto token sales via Echo's Sonar platform, the company later plans on expanding support to tokenized securities and real-world assets. Echo was founded by crypto trader Jordan Fish, widely known by his "Cobie" pseudonym. The platform has helped crypto projects raise more than $200 million since its launch two years ago. In May, Coinbase had struck a $2.9 billion deal for crypto options provider Deribit, plugging a gap in its derivatives portfolio and strengthening its international presence. (Reporting by Arasu Kannagi Basil in Bengaluru; Editing by Shilpi Majumdar)

(The article has been published through a syndicated feed. Except for the headline, the content has been published verbatim. Liability lies with original publisher.)

(Reuters) -AI startup Anthropic is in talks with Google to secure additional computing power valued in the high tens of billions of dollars, Bloomberg News reported on Tuesday, citing people familiar with the matter. The deal, which has not been finalized yet, includes Alphabet's Google providing cloud computing services to Anthropic, according to the report. Alphabet's stock gained 2.3% in trading after the bell. Google declined to comment, while Anthropic did not immediately respond to a request for comment. Anthropic, which counts Google and Amazon.com as its major investors, is the creator of the Claude chatbot and competes with OpenAI's ChatGPT for AI dominance. Anthropic is projecting to more than double and potentially nearly triple its annualized revenue run rate next year, fueled by the rapid adoption of its enterprise products, Reuters had exclusively reported last week. The company is on track to meet an internal goal of $9 billion in annual revenue run rate – a calculation of annual revenue extrapolated from the current sales pace – by the end of 2025. (Reporting by Juby Babu in Mexico City; Editing by Anil D'Silva)

(The article has been published through a syndicated feed. Except for the headline, the content has been published verbatim. Liability lies with original publisher.)

By Lisa Richwine LOS ANGELES (Reuters) -Netflix missed Wall Street's third-quarter earnings targets because of an unexpected expense from a Brazilian tax dispute while it offered a forecast a touch ahead of Wall Street projections for the rest of the year. The report failed to impress investors accustomed to fast-paced growth from the streaming video pioneer. Shares of Netflix, which had risen 39% this year ahead of the earnings release, fell 5.6% to $1,171.24 in after-hours trading on Tuesday. Netflix is seeking to expand in new areas such as advertising and video games after attracting more than 300 million customers around the world. It faces competition from YouTube, Amazon's Prime Video, Disney+ and others. The media business is facing major changes including the potential sale of industry titan Warner Bros Discovery and the rise of generative artificial intelligence with the ability to produce short videos. CNBC reported that Netflix was among the parties interested in examining Warner Bros' assets. Netflix Co-CEO Ted Sarandos, in response to analysts' questions about possible media consolidation, said the company "can and will be choosy" about acquisition targets. He said the company has no interest in owning legacy media networks but would evaluate opportunities to buy intellectual property. "Nothing is a must-have for us to meet our goals that we have for the business," Sarandos said. Co-CEO Greg Peters said he did not believe media industry consolidation would make things more challenging for Netflix. "Watching some of our competitors potentially get bigger via (mergers and acquisitions) does not change in and of itself, at least our view, the competitive landscape," Peters said. Netflix posted net income of $2.5 billion and diluted earnings-per-share of $5.87 for July through September, a period when the animated "K-Pop Demon Hunters" became the most-watched movie in Netflix history. Analysts had expected $3.0 billion and $6.97, respectively, according to LSEG. Revenue was even with forecasts, at $11.5 billion. Netflix reported an operating margin of 28% for the third quarter. Without the Brazilian tax expense of roughly $619 million, the margin would have exceeded the company's guidance of 31.5%, it said, adding that it did not expect the matter to have a material impact on future results. PP Foresight analyst Paolo Pescatore said he believed the tax issue weighed on Netflix shares. "All things considered, this was another robust quarter, despite a blip due to an unforeseen expense," Pescatore said. For the fourth quarter, Netflix forecast revenue of $11.96 billion, compared with Wall Street's projection of $11.90 billion. It projected diluted earnings-per-share a penny ahead of analysts' targets, at $5.45. For the third quarter, Netflix said it recorded its best ad sales quarter in history but did not disclose a number. "This gives the impression that the sustained revenue growth achieved this quarter, and forecasted for next quarter, will predominantly continue to come from subscription fees," eMarketer analyst Ross Benes said. Netflix will release the final season of one of its biggest hits, "Stranger Things," in November and December and stream two live National Football League games on Christmas. "We're finishing the year with good momentum and have an exciting Q4 slate," Netflix said in its quarterly letter to shareholders. Earlier this year, Netflix stopped reporting subscriber numbers and urged investors to focus on revenue and profit. It has expanded into video games and advertising, two areas that have contributed little to revenue so far, according to analysts and investors.  (Reporting by Lisa Richwine; Editing by Richard Chang)

(The article has been published through a syndicated feed. Except for the headline, the content has been published verbatim. Liability lies with original publisher.)

(Reuters) -TeamViewer said on Tuesday it expects full-year revenue to be at the lower end of its prior outlook of 778 million euros to 797 million euros ($907 million-$929.5 million), driven by weakness in its recently acquired 1E business. Standalone annual recurring revenue at 1E, a December 2024 acquisition, which develops products that detect IT issues, was below expectations due to "ongoing transformation and persistent macroeconomic challenges", the company said in a statement. In the third quarter, overall revenue grew 4% on a constant currency basis, while sales from 1E slipped 8%. In addition to the muted macroeconomic environment in Europe, "ongoing macro challenges" in the United States, 1E's traditionally strongest market, affected sales by slower customer decision-making and reduced deal volumes, it said. "Ongoing initiatives to turn around this part of the business will take time to materialize, affecting our short-term growth outlook," TeamViewer CFO Michael Wilkens said. Overall annual recurring revenue is now expected in the range of 780 million euros to 800 million euros, from 815 million euros to 840 million euros previously. Revenue growth expectations for 2026 have also slowed to 2%-6%. The company will implement more cost-cutting measures to partly offset the hit to revenues, it said. TeamViewer, however, increased its forecast for 2025 adjusted earnings before interest, taxes, depreciation and amortization margin to 44% from 43%, saying it managed to keep a lid on costs. ($1 = 0.8575 euros) (Reporting by Chandni Shah in Bengaluru, Chris Thomas in Mexico City; Editing by Alan Barona)

(The article has been published through a syndicated feed. Except for the headline, the content has been published verbatim. Liability lies with original publisher.)

By Olivia Le Poidevin GENEVA (Reuters) -Leading AI assistants misrepresent news content in nearly half their responses, according to new research published on Wednesday by the European Broadcasting Union (EBU) and the BBC. The international research studied 3,000 responses to questions about the news from leading artificial intelligence assistants – software applications that use AI to understand natural language commands to complete tasks for a user. It assessed AI assistants in 14 languages for accuracy, sourcing and ability to distinguish opinion versus fact, including ChatGPT, Copilot, Gemini and Perplexity. Overall, 45% of the AI responses studied contained at least one significant issue, with 81% having some form of problem, the research showed. Reuters has made contact with the companies to seek their comment on the findings. Gemini, Google's AI assistant, has stated previously on its website that it welcomes feedback so that it can continue to improve the platform and make it more helpful to users. OpenAI and Microsoft have previously said hallucinations – when an AI model generates incorrect or misleading information, often due to factors such as insufficient data – are an issue that they are seeking to resolve. Perplexity says on its website that one of its "Deep Research" modes has 93.9% accuracy in terms of factuality. SOURCING ERRORS A third of AI assistants' responses showed serious sourcing errors such as missing, misleading or incorrect attribution, according to the study. Some 72% of responses by Gemini, Google's AI assistant, had significant sourcing issues, compared to below 25% for all other assistants, it said. Issues of accuracy were found in 20% of responses from all AI assistants studied, including outdated information, it said. Examples cited by the study included Gemini incorrectly stating changes to a law on disposable vapes and ChatGPT reporting Pope Francis as the current Pope several months after his death. Twenty-two public-service media organisations from 18 countries including France, Germany, Spain, Ukraine, Britain and the United States took part in the study. With AI assistants increasingly replacing traditional search engines for news, public trust could be undermined, the EBU said. "When people don’t know what to trust, they end up trusting nothing at all, and that can deter democratic participation," EBU Media Director Jean Philip De Tender said in a statement. Some 7% of all online news consumers and 15% of those aged under 25 use AI assistants to get their news, according to the Reuters Institute’s Digital News Report 2025. The new report urged AI companies to be held accountable and to improve how their AI assistants respond to news-related queries. (Reporting by Olivia Le Poidevin, Editing by Timothy Heritage)

(The article has been published through a syndicated feed. Except for the headline, the content has been published verbatim. Liability lies with original publisher.)

By Jaspreet Singh and Aditya Soni (Reuters) -Meta has struck a $27 billion financing deal with Blue Owl Capital to fund its biggest data center project globally, as large technology companies race to build out the infrastructure needed to power their artificial intelligence ambitions. Tuesday's announcement marks Meta's largest-ever private capital deal. Under it, Meta will retain about 20% equity in the Louisiana project, with the majority owned by funds that alternative asset manager Blue Owl Capital manages. Blue Owl contributed roughly $7 billion in cash to the joint venture, with Meta receiving a one-time payout of about $3 billion. The planned data center in Richland Parish, Louisiana, known as Hyperion, is projected to deliver more than 2 gigawatts of compute capacity to support training of large language models, the technology behind tools such as ChatGPT and Google Gemini. Doug Ostrover and Marc Lipschultz, Co-CEOs of Blue Owl, called Hyperion "an ambitious project that reflects the scale and speed required to power the next generation of AI infrastructure." Major tech companies, including Alphabet, Amazon.com, Meta, Microsoft and CoreWeave, are on track to spend $400 billion on AI infrastructure this year, Morgan Stanley estimates. OpenAI, the startup at the heart of the AI boom, recently signed multiple deals that may cost over $1 trillion to secure about 26 gigawatts of computing capacity, enough to power roughly 20 million U.S. homes. Meta's finance chief, Susan Li, called Tuesday's deal "a bold step forward." The company has signed leases for the facility with a four-year initial term with an option to extend and expects the project to create more than 500 jobs once it goes online. "It definitely helps them mitigate risks at the expense of its ownership stake. Meta won't need to provide nearly as much capital and can look to finance other facilities or AI infrastructure," said Alvin Nguyen, senior analyst at Forrester. "It also minimizes the debt they are taking on for equipment and property in the event there is a bursting of the AI bubble." (Reporting by Jaspreet Singh and Aditya Soni in Bengaluru; Editing by Tasim Zahid and Alan Barona)

(The article has been published through a syndicated feed. Except for the headline, the content has been published verbatim. Liability lies with original publisher.)