Technology

Joe Tsai, Alibaba chairman: "The only way for Europe to achieve technological sovereignty is with open source AI"
Joe Tsai, chairman and co-founder of Alibaba, said at the Wave event in Italy that the only path for Europe to achieve technological sovereignty is open source AI: downloading the weights of an open model, running it on its own infrastructure and training it with internal data so that sensitive information does not leave the company. According to Tsai, Chinese AI companies share their innovation while US labs do not. The article notes that China and the US control around 90% of the world's AI computing infrastructure.
TechnologyInternationalXatakaOct 25AI
Colombian model Akima denounces use of her face to create AI avatars
Akima, a Colombian model and actress in Netflix's series 'One Hundred Years of Solitude', denounced that Mexican producer Mauricio Sierra used photographs of her face to create between six and seven avatars with artificial intelligence for campaigns by luxury brands in Europe, the United States and Latin America. According to her account, Sierra justified the characters by saying he 'couldn't find these faces in the world'. The model described the situation as a racial expropriation of her identity and announced she will explore legal action from the United States and Europe.
TechnologyInternationalInfobaeOct 25AI
Apollo: AI shows no return for the 493 non-tech S&P 500 companies
Apollo Global Management chief economist Torsten Slok reviewed the accounts of the 493 non-tech companies in the S&P 500 and found no sign that AI is improving their margins or revenue per worker. According to Bloomberg, revenue per worker is rising in the Magnificent Seven but flat in the so-called S&P 493. Slok also notes that token spending is starting to be seen as a cost center pressuring margins.
TechnologyInternationalEl Ecosistema StartupOct 25AI
Users report cracked screens on the iPhone Duo, according to complaints on Reddit
Since Friday, iPhone Duo users have been reporting on Reddit that the screen cracks around the central hinge line, leaving a horizontal block of dead pixels with intense white light. The specialist outlet 9to5Mac has picked up the cases, which include a unit on display at an Apple Store in Toronto. Some accounts place the breakage a day after purchase or during initial setup. There is no official explanation from Apple; one circulating theory points to a batch with a misaligned hinge.
TechnologyInternationalADSLZoneOct 25AI
NHTSA proposes optional cyan lights to identify autonomous vehicles
The NHTSA, the U.S. federal road safety regulator, published a proposal to rewrite lighting standard FMVSS No. 108 to adapt it to driverless vehicles. The centerpiece is cyan marker lights, visible from the front, rear and both sides, that would indicate when the automated driving system is operating. They would be optional: if a manufacturer installs them, it must meet federal standards for color, brightness, location and operation. The proposal also removes the requirement for manual controls for turn signals and high beams in vehicles designed to operate autonomously.
TechnologyInternationalEl Ecosistema StartupOct 25AI
Valeo CEO: cars can be autonomous without €100,000 worth of sensors
Christophe Périllat, CEO of Valeo, said in an interview with Automotive News that "cars can be autonomous without carrying €100,000 worth of sensors on the roof". The executive shares Elon Musk's thesis on camera- and AI-based autonomous driving, and argues that driving depends not only on perception but also on learning. According to Périllat, "in 10 or 15 years we will see all cars become autonomous".
TechnologyInternationalCar and DriverOct 25AI
Apollo chief economist says AI has delivered "zero return" to US companies outside the tech sector
Torsten Slok, chief economist at Apollo Global Management, argues that artificial intelligence has delivered "zero return" to the US corporate economy outside the tech sector that sells it, according to Bloomberg. His argument compares the "Magnificent Seven" of the S&P 500 with the other 493 companies: revenue per worker is rising at the former and stagnant at the rest, and margins show no improvement either. Slok warns that spending on tokens is starting to be seen as an added cost and that AI's macroeconomic effects will be limited while it remains concentrated in a small slice of the economy.
TechnologyInternationalBioBioChileOct 25AI
Affective bonds with AI: a study documents symbolic weddings and emotional dependence
A study by CSIC and the Universitat Politècnica de Valencia, led by researcher José Such, documented through interviews with 17 people affective relationships with AI assistants such as ChatGPT, Character.AI or Replika, including symbolic weddings, regular dates and simulated pregnancies. Specialists note that the brain does not need a human stimulus to activate reward and attachment mechanisms, and warn about emotional dependence and the replacement of human bonds.
TechnologyInternationalInfobaeOct 25AI
Nvidia weighs charging about US$2,000 per GPU per year for software that runs AI agents
Nvidia CEO Jensen Huang said at GTC Berlin on October 22 that the company is weighing licensing its software for running AI agents in data centers at around US$2,000 per GPU per year. According to the calculation described by FourWeekMBA, five million GPUs under paid license would amount to US$10 billion a year. Nvidia did not confirm how many GPUs would be subject to the fee, which applications the license covers or when it would take effect. Its AI Enterprise platform is listed at US$4,500 per GPU per year.
TechnologyInternationalEl Ecosistema StartupOct 25AI
Anthropic releases sandbox-runtime, an open-source tool to isolate AI agents
Anthropic released sandbox-runtime, an open-source tool that confines any process within file and network limits without spinning up a container. It is distributed on npm as @anthropic-ai/sandbox-runtime, under the Apache-2.0 license, and requires Node.js 22.12 or later. It uses the host operating system's isolation capabilities and applies restrictions to the entire process tree. By default it allows reads, blocks writes unless authorized, and denies network access with exceptions through an allowed-domain list. Anthropic reports an 84% reduction in permission prompts in its internal use of Claude Code.
TechnologyInternationalEl Ecosistema StartupOct 25AI
Sweden bans ChatGPT and Claude in public administration, leaving only Microsoft Copilot
Starting Monday, October 26, employees at Swedish ministries and state buildings will no longer be able to use ChatGPT or Claude on their work devices. The government made the decision citing security failures and left Microsoft Copilot as the only authorized alternative. Mikael Wettercrantz, head of digital strategy for the government offices network, warned about the risks of entering government information into platforms without a legal framework, which send data to external infrastructure and, under US law, can be handed to intelligence or judicial agencies.
TechnologyInternationalMARCAOct 25AI
MrBeast invests $250,000 in Ball 2, an AI-powered ball that creates games
Jimmy Donaldson, MrBeast, made his debut as a guest shark on the season 18 premiere of 'Shark Tank' and took 55% of Sperry Labs, the Pennsylvania company behind Ball 2, for $250,000. The founders had asked for $200,000 for 10%, valuing the company at $2 million; the final deal puts it at about $455,000. Ball 2 is a foam ball with sensors and Bluetooth that tracks throws and sends them to an app, where AI writes the game rules.
TechnologyInternationalXatakaOct 25AI
San Francisco Chronicle editor: "Artificial intelligence is going to be cheaper than what we journalists charge"
Emilio García-Ruiz, editor-in-chief of the San Francisco Chronicle, said artificial intelligence "is going to be at a much lower price than what we have to charge" and flagged as the main risk that AI companies take their content and build a better experience for readers. He compared it to music moving to digital, when "good enough" prevailed. He also said journalists must be "influencers" on their beats and that in an ideal world 80% of the newsroom would work in digital and 20% in print.
TechnologyInternationalPortafolioOct 25AI
Nvidia adds BBVA, Santander, TIM and Bolt to its AI alliance network
Nvidia presented a series of agreements with technology, banking, telecommunications and automotive companies at its GTC conference in Berlin. BBVA and Santander will explore with Nvidia the use of the OpenShell Reference Architecture, an open-source governance framework for AI agents in critical systems. Siemens is also involved in chip design, TIM and Nokia in telecommunications, and Bolt, Volvo Cars and Wayve in autonomous driving. Nvidia provides a common technical base and each partner develops its own application.
TechnologyInternationalEl Ecosistema StartupOct 25AI
Mexican consumers turn to AI to decide what to buy, NielsenIQ says
The Consumer Outlook 2027 study by NielsenIQ (NIQ) indicates that Mexican consumers have the highest intention to increase spending in Latin America, though they increasingly analyze whether their purchases are worth it. 38% of Mexicans believe their financial situation improved versus the previous year, 37% see it as stable and 25% say it worsened. Mexico ranks third in the region in the use of artificial intelligence during the purchase process, behind Brazil and Chile.
TechnologyInternationalTiempo La Noticia DigitalOct 25AI
Installing the XUPER TV APK on Android exposes users to password theft and ads that can't be closed
Installing the XUPER TV APK outside Google Play Store exposes users to risks beyond content: password theft, access to personal files, persistent advertising and the download of additional programs without security guarantees. According to ESET, which analyzed the Magis TV APK, now XUPER TV, alternative download platforms can distribute malware, spyware and tools designed to collect personal information. The problem lies not only in the app but in the installation method, which falls outside official store review controls.
TechnologyInternationalInfobaeOct 25AI
96% of travelers would switch their favorite brand if AI offered a faster process, Accenture says
An Accenture study based on a survey of more than 5,000 travelers in 12 markets indicates that 96% would abandon their preferred travel brand if another provider offered a significantly easier and faster experience, and that 91% would pay more to reduce the time and effort of planning a trip. 94% describe the booking process as complex and more than 83% abandon a purchase due to payment problems. Agentic AI is beginning to influence the discovery and choice of airlines and hotels.
TechnologyInternationalInfobaeOct 25AI
Farol, the open source tool that maps a pull request's risk before approving it
A Harvard study of more than 700 companies and 300 million work events found that AI agents increase generated lines of code by 30%, commits by 20% and pull requests by 23%, without significantly changing the pace of feature resolution. Winnin released farol, a free open source CLI and GitHub Action (MIT license) that analyzes the full program with real compilers and delivers a risk verdict before the reviewer opens the diff.
TechnologyInternationalEl Ecosistema StartupOct 25AI
Bernstein: blocking AI agents is defensible for Amazon, Uber and DoorDash for now
In an October 23 note, Bernstein analysts argue that large online aggregators — Amazon, Uber and DoorDash — may be justified in blocking external AI agents for now, because the technology threatens their business models. The most concrete case is Muse, Meta's agent, which Amazon blocked from its store, alleging it did not identify itself and appeared to capture customer credentials. Bernstein outlines a two-speed scenario: dominant platforms have few incentives to speed up adoption, while challengers have everything to gain by integrating.
TechnologyInternationalEl Ecosistema StartupOct 25AI
Mesh WiFi dropouts despite perfect coverage point to 802.11k, v and r standards
Having wifi coverage throughout the home does not guarantee an uninterrupted connection. In Mesh networks, brief dropouts can stem from roaming between nodes: it is the client device that decides when to switch access point. Three standards play a role: 802.11k, which informs about nearby nodes; 802.11v, which suggests the most suitable one; and 802.11r or Fast BSS Transition, which speeds up authentication. Incompatibilities or incorrect settings can cause interruptions in video calls or online games.
TechnologyInternationalADSLZoneOct 25AI