विज्ञान

By Elvira Pollina and Giulio Piovaccari TURIN (Reuters) -Amazon founder Jeff Bezos predicted on Friday gigawatt-scale data centres will be built in space within the next 10 to 20 years and that continuously available solar energy meant they would eventually outperform those based on Earth. Speaking at the Italian Tech Week in Turin, Bezos also compared the surge in artificial intelligence to the internet boom of the early 2000s, urging optimism despite the risk of speculative bubbles. The concept of orbital data centres has gained traction among tech giants as those on Earth have driven up demand for electricity and water to cool their servers. "These giant training clusters, those will be better built in space, because we have solar power there, 24/7. There are no clouds and no rain, no weather," Bezos said in a public conversation with Ferrari and Stellantis Chairman John Elkann. "We will be able to beat the cost of terrestrial data centres in space in the next couple of decades." Bezos said the shift to space infrastructure is part of a broader trend of using space to improve life on Earth. "It's already happened with weather and communication satellites," he said. "The next step is data centres, then other kinds of manufacturing." Hosting data centres in space has its own challenges, including the difficulty of maintenance and carrying out upgrades and the cost of launching rockets, as well as the risk the launches may fail. The executive chair of Amazon said the AI wave shares traits with the dot-com era, when massive hype was followed by a crash. "We should be extremely optimistic that the societal and beneficial consequences of AI, like we had with internet 25 years ago, are for real and there to stay," he said. "It is important to decorrelate the potential bubbles and their bursting consequences that might or might not happen from the actual reality," Bezos said, adding that the benefits of AI were expected "to be broadly diffused and it will go everywhere". (Reporting by Elvira Pollina and Giulio Piovaccari; Writing by Elvira Pollina; Editing by Emelia Sithole-Matarise and Barbara Lewis)

(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) -OpenAI CEO Sam Altman has embarked on a global fundraising initiative spanning East Asia and the Middle East, seeking financing and manufacturing partners to meet the company's demand for its computing capacity, the Wall Street Journal reported on Saturday, citing sources familiar with the meetings. Altman has held discussions since late September in Taiwan, South Korea, and Japan with major suppliers such as Taiwan Semiconductor Manufacturing, Foxconn, Samsung, and SK Hynix to boost AI chip production and secure top-priority orders for OpenAI, the report said. Altman was pushing these companies to increase production capacity and give priority to OpenAI's orders, the newspaper added. Reuters could not immediately verify the report. Microsoft, TSMC, Foxconn, Samsung and SK Hynix couldn't immediately respond to requests for comment outside regular business hours. Earlier this week, Samsung Electronics and SK Hynix had signed letters of intent to supply memory chips for OpenAI's data centers. Altman has planned to visit investors in the United Arab Emirates to raise money to help fund OpenAI's infrastructure expansion and research, WSJ said. OpenAI recently told its investors and business partners that it was likely to spend around $16 billion in renting computing servers this year, and that the expenditure could rise to around $400 billion in 2029, the report added, citing people familiar with the matter. (Reporting by Dheeraj Kumar in Bengaluru; Editing by Alexandra Hudson)

(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) -OpenAI CEO Sam Altman has embarked on a global fundraising initiative spanning East Asia and the Middle East, seeking financing and manufacturing partners to meet the company's demand for its computing capacity, the Wall Street Journal reported on Saturday, citing sources familiar with the meetings. Altman has held discussions since late September in Taiwan, South Korea, and Japan with major suppliers such as Taiwan Semiconductor Manufacturing, Foxconn, Samsung, and SK Hynix to boost AI chip production and secure top-priority orders for OpenAI, the report said. Altman was pushing these companies to increase production capacity and give priority to OpenAI's orders, the newspaper added. Reuters could not immediately verify the report. Microsoft, TSMC, Foxconn, Samsung and SK Hynix couldn't immediately respond to requests for comment outside regular business hours. Earlier this week, Samsung Electronics and SK Hynix had signed letters of intent to supply memory chips for OpenAI's data centers. Altman has planned to visit investors in the United Arab Emirates to raise money to help fund OpenAI's infrastructure expansion and research, WSJ said. OpenAI recently told its investors and business partners that it was likely to spend around $16 billion in renting computing servers this year, and that the expenditure could rise to around $400 billion in 2029, the report added, citing people familiar with the matter. (Reporting by Dheeraj Kumar in Bengaluru; Editing by Alexandra Hudson)

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

It's not only for health reasons that we are being told to eat less red meat. The climate and environment are under massive strain from our meat-heavy diet. There's still a long way to go when it comes to changing our food systems, new research shows. Potsdam, Germany (dpa) – Meat is murder – but not only of animals. Our dependence on environmentally unsustainable foods like red meat is driving a climate crisis that is claiming lives in floods, storms and other natural disasters made more likely by climate change, and the authors of a new study are calling on us to change. Global food production is responsible for around 30% of global greenhouse gas emissions, according to a damning new report by the EAT-Lancet Commission, accusing humanity of failing to address a critical issue in the climate crisis. Even if we manage to transition global energy away from fossil fuels, our current food systems will prevent us from limiting global warming to 1.5 degrees Celsius, the group warns. At the same time, the researchers note that while wealthy nations consume more than their share of carbon-intensive foods, billions of people lack access to healthy nutrition. "More than half of the world's population struggles to access healthy diets, leading to devastating consequences for public health, social equity, and the environment," the authors write. The commission, an international panel of experts established by the non-governmental organisation EAT and the journal The Lancet, aims to develop strategies for healthy human diets while protecting the planet. For a stable climate and healthy planet "Transforming food systems is a major ecological and social challenge, but it is a prerequisite for having a chance to return to a stable climate system and a healthy planet," says Johan Rockström, director of the Potsdam Institute for Climate Impact Research (PIK) and co-chair of the EAT-Lancet Commission. "Our eating habits can save lives, drastically reduce emissions, slow biodiversity loss and contribute to greater equity." The report says that if humanity can shift to what's known as the Planetary Health Diet, we could save approximately 15 million lives annually that are prematurely ended by diet-related diseases. The diet closely resembles common nutritional recommendations: plenty of fruits and vegetables, whole grains, nuts and legumes, along with three to four eggs per week, fish, poultry and only minimal meat from cattle, sheep and goats. Ruminants emit significant greenhouse gases The consumption of red meat not only poses a long-term cancer risk but also contributes significantly to greenhouse gas emissions. During digestion, animals like cows, sheep and deer produce large amounts of methane, a potent greenhouse gas. According to the report, 53% of non-CO2 greenhouse gases from agriculture come from ruminants. If people worldwide adopted healthier diets, emissions from the food sector could be reduced by 15%. Current eating habits, experts say, often lack fruits, vegetables, nuts, legumes and whole grains, while meat, dairy products, fats, sugar and highly processed foods are consumed in excess. The goal, the group says, is not a one-size-fits-all diet. The Planetary Health Diet is flexible and compatible with a wide variety of foods, cultures, dietary habits, traditions and individual preferences. However, significant changes would be required in how, where, and what food is produced. Some sectors would need to shrink — for example, red meat production would need to decrease by about one-third. Other areas, such as fruit, vegetable, and nut production, would need to expand by nearly two-thirds compared to 2020 production levels. Other proposed solutions include reducing food loss and waste along the supply chain and promoting agricultural practices that increase productivity while minimising environmental impacts. This includes less invasive soil cultivation methods compared to deep ploughing. A climate-neutral food system is theoretically possible Unprecedented investments and efforts in all these areas could potentially lead to a climate-neutral food system, the report states. The findings build on the analysis of a previous commission presented in 2019. Currently, food production is the single largest driver of exceeding planetary boundaries, which include not only climate change but also disruptions to the nitrogen and phosphorus cycles and ocean acidification. "Obesity rates continue to rise globally, and the pressure exerted by food systems on planetary boundaries shows no signs of abating," the authors write. "The report provides the clearest guidance yet on how to feed a growing global population without exceeding the Earth's safe operating space as defined by planetary boundaries," says Rockström, lead author of thereport , published in October. The experts also highlight social aspects, calling for fair wages and safe working conditions for all food sector workers, as well as greater inclusion of small-scale farmers, Indigenous peoples, women and other often underrepresented groups in decision-making processes. The following information is not intended for publication dpa fm yyzz n1 zlw kll coh

(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 Mrinmay Dey (Reuters) -ChatGPT creator OpenAI will soon introduce controls allowing the owners of content rights to dictate how their characters are used in its AI video-generating tool Sora and plans to share revenue with those who permit such use. The artificial intelligence company will give rights holders "more granular control over generation of characters," Chief Executive Sam Altman posted on his blog on Friday. Altman said options for copyright owners, such as television and movie studios, will include being able to block the use of their characters. Scrutiny is growing over AI-generated content and its impact on intellectual property rights, as companies navigate how to balance innovation with fair compensation for creators. OpenAI launched Sora this week as a standalone app, initially available in the United States and Canada. Videos in the app can be up to 10 seconds long. The app, which swiftly rose in popularity, lets users create and share AI videos that can be spun from copyrighted content and shared to social media-like streams. Its copyright policy is expected to stir tensions in Hollywood. At least one major studio, Disney, has opted out of having their material appear in the app, people familiar with the matter have told Reuters. OpenAI also plans to introduce a revenue-sharing model for copyright holders who permit their characters to be generated by users, Altman wrote. He said users are creating significantly more video content than expected, often for niche audiences, prompting the need for a monetization strategy. Altman acknowledged that the revenue-sharing framework "will take some trial and error to figure out," but said implementation would begin soon as OpenAI intends to test various approaches within Sora, before rolling out a consistent model across its broader product suite. Microsoft-backed OpenAI launched a Sora model for public use last year, expanding its foray into multimodal AI technologies and competing with similar text-to-video tools from Meta and Alphabet's Google. Meta recently unveiled Vibes, a platform where users can create and share short-form, AI-generated videos. (Reporting by Mrinmay Dey in Bengaluru; Editing by William Mallard)

(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 Debora Ely (Reuters) -A group of suspected scammers in Brazil using Instagram ads featuring deepfakes of supermodel Gisele Bundchen and other celebrities appears to have made millions of dollars through online fraud, according to police investigators. Authorities arrested four suspects tied to the scheme this week and froze assets across five states in an investigation that found over 20 million reais ($3.9 million) in suspicious funds identified by federal anti-money laundering agency COAF. The operation marks one of the first attempts in Brazil to counter the growing use of artificial intelligence tools to manipulate images and videos of celebrities for online scams. Brazil's Supreme Court ruled in June that social media platforms can be held liable for criminal ads posted by users if they fail to act swiftly to remove the content, even without a court order.  Instagram owner Meta said that its policies forbid "ads that deceptively use public figures to try to scam people," and said it removes such ads "when detected." The company said in a statement that it has "specialized systems to detect celeb-bait, invests heavily in trained review teams, shares tips on avoiding scams and offers tools to report potential violations." A spokesperson for Bundchen's team advised consumers to exercise caution regarding unusual discounts in advertisements featuring celebrities, verify the authenticity of offers through official brand or celebrity channels, and report potential scams to authorities. The probe announced this week by police investigators in the southern state of Rio Grande do Sul began in August 2024 after a victim reported being misled by an Instagram ad showing an altered video of Bundchen promoting a skincare product.  Another ad used the supermodel's likeness to promise a giveaway of suitcases, with buyers paying shipping fees for items that never arrived.  "We identified that the criminal group carried out a series of other scams, involving deepfakes of other celebrities and false betting platforms," Eibert Moreira Neto, head of Rio Grande do Sul's cybercrime unit, told Reuters.  Investigators are pursuing charges of money laundering and online fraud. They noted that most victims lost small amounts, usually under 100 reais ($19), and did not report the crimes.  "That created a perverse situation in which the criminals enjoyed a kind of 'statistical immunity.' They knew most people would not report them, so they operated at scale without fear," Isadora Galian, from the cybercrime unit, said in a statement.  (Reporting by Debora Ely; Additional reporting by Bernardo Barbosa; Editing by Brad Haynes, Aurora Ellis and Leslie Adler)

(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 Mike Stone WASHINGTON (Reuters) -The much-needed modernization of the U.S. Army's battlefield communications network being undertaken by Anduril, Palantir and others is rife with "fundamental security" problems and vulnerabilities, and should be treated as a "very high risk," according to a recent internal Army memo. The two Silicon Valley companies, led by allies of U.S. President Donald Trump, have gained access to the Pentagon's lucrative flow of contracts on the promise of quickly providing less expensive and more sophisticated weapons than the Pentagon's longstanding arms providers.   Military drone and software maker Anduril boasted it had a prototype of the NGC2 communications platform working during a battlefield test just eight weeks after winning the contract award. But the September 5 memo provides fodder for critics who argue that Silicon Valley’s move-fast, break-things ethos may not be the best approach for vital military equipment.  The memo from the Army's chief technology officer about the NGC2 platform that connects soldiers, sensors, vehicles and commanders with real-time data paints a bleak security picture of the initial product. "We cannot control who sees what, we cannot see what users are doing, and we cannot verify that the software itself is secure," the memo says. Those concerns have been addressed already as part of the "normal process" of development, Anduril said. "The recent report reflects an outdated snapshot, not the current state of the program," the company said in a statement emailed to Reuters.  A Palantir spokesperson said, "No vulnerabilities were found in the Palantir platform." However, the Army internal memo written by Gabriele Chiulli, the Army chief technology officer authorizing official on the NGC2 prototype, said, "Given the current security posture of the platform and the hosted 3rd party applications the likelihood of an adversary gaining persistent undetectable access to the platform requires the system be treated as very high risk." Palantir stock closed down 7.5% on Friday. Anduril is not publicly traded, although company founder Palmer Luckey has said a public offering is planned. A September 30 article on Anduril’s website touted the NGC2 system’s performance during a live-fire exercise in Fort Carson, Colorado: "Soldiers fired 26 live missions with M777 howitzers on Fort Carson's live-fire ranges, running AXS side-by-side with legacy crews. The contrast was visible: one team struggling with delays, the other firing digitally in seconds." Leonel Garciga, Army chief information officer and Chiulli's supervisor, told Reuters in an interview on Friday that frank vendor communication is important. "I think there's only one application that still has some vulnerabilities that they were working on," Garciga said, adding that many issues were fixed within a matter of weeks and days. The Army CTO's assessment, seen by Reuters and first reported by Breaking Defense, comes just months after Anduril was awarded a $100 million contract to create a prototype of NGC2 with partners including Palantir, Microsoft and several smaller contractors. The memo said the system allows any authorized user to access all applications and data regardless of their clearance level or operational need. As a result, "Any user can potentially access and misuse sensitive" classified information, the memo states, with no logging to track their actions. Other deficiencies highlighted in the memo include the hosting of third-party applications that have not undergone Army security assessments. One application revealed 25 high-severity code vulnerabilities. Three additional applications under review each contain over 200 vulnerabilities requiring assessment, according to the document. Garciga said that next week the Palantir Federal Cloud Service, which the battlefield system relies on, could be approved by the Army to deploy software updates more quickly after receiving a milestone permission known as "continuous authority to operate." Palantir and Anduril are leading a new wave of defense firms aiming to transform the U.S. military with cutting-edge technologies, including drones, artificial intelligence and autonomous systems. Palantir, a data analytics firm known for its work with intelligence agencies, and Anduril have seen their valuations surge under Trump’s presidency, as Washington increasingly turns to Silicon Valley-style innovation to counter rising threats. Anduril recently signed a $159 million deal for an initial prototyping period to develop a night vision and mixed-reality system as part of the Soldier Borne Mission Command (formerly IVAS Next) program. Palantir signed a $480 million contract for Maven, an artificial intelligence tool that sifts through images and sensor data to provide battlefield analysis. (Reporting by Mike Stone; editing by Chris Sanders, Sergio Non and Leslie Adler)

(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 Debora Ely (Reuters) -A group of suspected scammers in Brazil using Instagram ads featuring deepfakes of supermodel Gisele Bundchen and other celebrities appears to have made millions of dollars through online fraud, according to police investigators. Authorities arrested four suspects tied to the scheme this week and froze assets across five states in an investigation that found over 20 million reais ($3.9 million) in suspicious funds identified by federal anti-money laundering agency COAF. The operation marks one of the first attempts in Brazil to counter the growing use of artificial intelligence tools to manipulate images and videos of celebrities for online scams. Brazil's Supreme Court ruled in June that social media platforms can be held liable for criminal ads posted by users if they fail to act swiftly to remove the content, even without a court order.  Instagram owner Meta said that its policies forbid "ads that deceptively use public figures to try to scam people," and said it removes such ads "when detected." The company said in a statement that it has "specialized systems to detect celeb-bait, invests heavily in trained review teams, shares tips on avoiding scams and offers tools to report potential violations." A spokesperson for Bundchen's team advised consumers to exercise caution regarding unusual discounts in advertisements featuring celebrities, verify the authenticity of offers through official brand or celebrity channels, and report potential scams to authorities. The probe announced this week by police investigators in the southern state of Rio Grande do Sul began in August 2024 after a victim reported being misled by an Instagram ad showing an altered video of Bundchen promoting a skincare product.  Another ad used the supermodel's likeness to promise a giveaway of suitcases, with buyers paying shipping fees for items that never arrived.  "We identified that the criminal group carried out a series of other scams, involving deepfakes of other celebrities and false betting platforms," Eibert Moreira Neto, head of Rio Grande do Sul's cybercrime unit, told Reuters.  Investigators are pursuing charges of money laundering and online fraud. They noted that most victims lost small amounts, usually under 100 reais ($19), and did not report the crimes.  "That created a perverse situation in which the criminals enjoyed a kind of 'statistical immunity.' They knew most people would not report them, so they operated at scale without fear," Isadora Galian, from the cybercrime unit, said in a statement.  (Reporting by Debora Ely; Additional reporting by Bernardo Barbosa; Editing by Brad Haynes, Aurora Ellis and Leslie Adler)

(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 Prakhar Srivastava (Reuters) -AI chip startup Cerebras Systems, which competes with industry leader Nvidia in the booming AI chip market, on Friday filed to withdraw its plans for an initial public offering in the United States, effective immediately. Its withdrawal comes as U.S. IPO activity picks up in recent months, reversing an earlier slowdown caused by trade-policy uncertainty, with recent listings, such as that of data center real estate investment trust Fermi, getting warm receptions amid surging investor enthusiasm for AI-related stocks. "Given that Cerebras just very recently completed a sizeable fund raise, it is of no surprise that they are holding off to pursue the IPO at this time," said Josef Schuster, CEO of IPO research firm IPOX. Cerebras had on Tuesday said it raised $1.1 billion in a funding round led by Fidelity Management & Research and Atreides Management, valuing the company at $8.1 billion. The round added investors Tiger Global, Valor Equity Partners and 1789 Capital, the fund in which U.S. President Donald Trump's son is a partner. But CEO Andrew Feldman said at the time the company still plans to hold a public offering. Last year, Cerebras filed for an initial public offering on the Nasdaq. The company’s highly anticipated listing was delayed by a U.S. national security review of a $335 million investment by G42, an Abu Dhabi-based cloud computing and AI company. "We believe this is more a company-specific strategic decision and does not tell us anything about the state of U.S. IPO sentiment, which we view as exceptionally strong," Schuster added. Sunnyvale, California-based Cerebras Systems makes high-performance AI chips and systems designed to speed up training and running large AI models. (Reporting by Prakhar Srivastava in Bengaluru; Editing by Alan Barona and Sahal Muhammed)

(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 Prakhar Srivastava (Reuters) -AI chip startup Cerebras Systems, which competes with industry leader Nvidia in the booming AI chip market, on Friday filed to withdraw its plans for an initial public offering in the United States, effective immediately. Its withdrawal comes as U.S. IPO activity picks up in recent months, reversing an earlier slowdown caused by trade-policy uncertainty, with recent listings drawing strong investor demand. Cerebras on Tuesday said it raised $1.1 billion in a funding round led by Fidelity Management & Research and Atreides Management, valuing the company at $8.1 billion. "Given that Cerebras just very recently completed a sizeable fund raise, it is of no surprise that they are holding off to pursue the IPO at this time," said Josef Schuster, CEO of IPO research firm IPOX. Last year, Cerebras filed for an initial public offering on the Nasdaq. The company’s highly anticipated listing was delayed by a U.S. national security review of a $335 million investment by G42, an Abu Dhabi-based cloud computing and AI company. "We believe this is more a company-specific strategic decision and does not tell us anything about the state of U.S. IPO sentiment, which we view as exceptionally strong," Schuster added. Sunnyvale, California-based Cerebras Systems makes high-performance AI chips and systems designed to speed up training and running large AI models, offering faster and cheaper alternatives to traditional GPUs. (Reporting by Prakhar Srivastava in Bengaluru; 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.)