Technology

Block, Jack Dorsey's company, unveils Dessa, a voice model trained only on human conversations
Block, Jack Dorsey's financial services and digital payments company formerly known as Square, presented Dessa, a voice model trained from scratch exclusively on real conversations. It is a full-duplex system, able to speak and listen at the same time, that reproduces laughter, pauses and crying. The company says it is the first AI with which a realistic live conversation can be held and will give early access to AI safety researchers to identify risks.
TechnologyInternationalReason WhyOct 26AI
AI pushes CIOs to the center of corporate boardrooms
61% of Spanish companies already use artificial intelligence in their daily operations, seven points above the European average according to an Amazon Web Services report. In a debate organized by EXPANSIÓN with Inetum and BMC Helix, technology leaders from Veiasa, Enagás, Bankinter, Asisa and BMC Helix discussed how to scale AI without increasing operational complexity or technical debt, and agreed on the need for governance, return metrics and digital literacy.
TechnologyInternationalExpansiónOct 26AI
Apollo report: 98% of S&P 500 companies see no profitability from AI
A report by investment firm Apollo, prepared by its chief economist Torsten Slok, says AI has not delivered tangible financial returns to S&P 500 companies outside the technology sphere. While revenue per employee and margins rise at the Magnificent Seven, both metrics remain stagnant in the S&P 493. The analysis adds that there are no measurable records of AI boosting overall productivity in the United States and that token spending is becoming an operating cost.
TechnologyInternationalGeekneticOct 26AI
Sam Altman warns AI poses a cybersecurity risk that is not yet fully grasped
OpenAI CEO Sam Altman said he cannot find words "dramatic enough" to describe the problem artificial intelligence poses to cybersecurity. According to the report, Altman fears traditional protections will be ineffective against autonomous AI agents and that isolating them in controlled environments will stop working. The text cites as warning signs a wave of attacks on PHP systems and the case of OpenAI agents that executed code on 41 production servers.
TechnologyInternationalXataka MóvilOct 26AI
Apple set to refresh the iMac with M6 chip this week, according to leaks
Apple will unveil new products this Wednesday, October 28, and according to leaks reported by Applesfera they would include an iMac refresh with the M6 chip, Apple's first built on a 2-nanometer process. The computer would keep the 23.5-inch IPS panel with 4.5K resolution and the design introduced in 2021, with possible new colors. Memory options of 16, 24 and 32 GB and storage from 256 GB to 2 TB are expected. Leaker Mark Gurman points to a sale date of November 6.
TechnologyInternationalApplesferaOct 26AI
Rio Tinto studies a 360 MWh brick thermal battery for its alumina refinery in Australia
Rio Tinto and US-based Rondo Energy are designing a 360 MWh thermal battery for the Yarwun alumina refinery near Gladstone (Queensland, Australia). The system heats refractory bricks with electricity, mainly solar, and delivers heat of up to 1,000 °C to generate industrial steam. According to Rio Tinto, it would cover about 5% of the plant's steam needs and avoid some 67,000 tonnes of CO2 per year. The Australian Renewable Energy Agency (ARENA) is funding the feasibility study.
TechnologyInternationalHíbridos y EléctricosOct 26AI
UNICEF warns 20 million children suffered technology-facilitated sexual abuse in one year
A report by UNICEF with INTERPOL and ECPAT International, titled "Through Children's Eyes", estimates that 20 million children and adolescents aged 12 to 17 in 21 countries suffered at least one form of technology-facilitated sexual exploitation or abuse in a single year. In the four Latin American and Caribbean countries included, 18% of adolescents who use the internet were victims. More than half of the cases occur on social media and nearly half of the victims told no one.
TechnologyInternationalInfobaeOct 26AI
A Basque engineer reveals his intern pay at Lenovo's AI lab in the US: $5,185 net per month
Siqiang Wu, a graduate in Industrial Technology Engineering from the University of the Basque Country, recounts his experience as an intern at Lenovo's Artificial Intelligence Lab in the United States. He worked 40 hours a week from June 3 to November 15, 2024, researching large language models, and then moved to full time. In five and a half months he earned more than $33,000 gross and about $28,500 net, plus $2,500 in relocation aid. He compares that pay with internships in Spain.
TechnologyInternationalEl Diario VascoOct 26AI
Toll system failure in Japan left 106 toll booths out of service for 38 hours
NEXCO Central's electronic toll collection system failed in the early hours of April 6, 2025, leaving 106 toll booths unusable for 38 hours across 17 routes, including those leading to Tokyo. The company raised the barriers and enabled a web form for deferred payment. About 960,000 vehicles passed through those toll booths and 3.8% requested to pay. The failure originated in a software update and meant 1.2 billion yen went uncollected.
TechnologyInternationalXatakaOct 26AI
One in six female MEPs has been targeted by AI-generated fake sexual content
Research by Agora Digital Transformation published in October says 46 female MEPs, roughly one in six and from different political groups, have been targeted with AI-generated fake sexual images, such as photos placing their real face on someone else's nude body or apps that simulate undressing them. The Greens are calling for enforcement of the European ban on apps that allow 'undressing', a debate scheduled for December 2. The text also cites cases involving Rosalía, Laura Escanes, Taylor Swift and Scarlett Johansson.
TechnologyInternationalLa VanguardiaOct 26AI
Study finds "early evidence" of higher unemployment in occupations most exposed to AI
A report by the Institut d'Economia de Barcelona (IEB) of the Universitat de Barcelona, using SEPE data cross-referenced with companies' activity profiles, finds "early evidence" that AI is associated with higher unemployment in the most exposed occupations. A ten-percentage-point increase in AI exposure is linked to a 1.2% rise in unemployment between ChatGPT's launch in 2022 and 2024, and 1.7% between 2024 and 2026. The authors do not observe a contraction in hiring at those companies.
TechnologyInternationalLa VanguardiaOct 26AI
AI reshapes threats as Barcelona Cybersecurity Congress gathers 150 experts and 111 companies
INCIBE recorded 122,223 incidents in Spain in 2025, 26% more than the previous year, and reported 237,028 vulnerable systems. In Catalonia, the Agència de Ciberseguretat handled 6,544 unique cyber incidents, nearly double that of 2024, and detected more than 9.1 billion attack attempts, of which more than 7 billion were blocked automatically. With AI changing how attacks are carried out, the seventh edition of the Barcelona Cybersecurity Congress will bring together more than 150 speakers and 111 companies at Fira de Barcelona from November 3 to 5.
TechnologyInternationalLa VanguardiaOct 25AI
A Peruvian novel written 99% by AI sparks debate over literary authorship
The novel "Donde el agua trabaja", by Peruvian author David Silva Rojas, contains 99% text generated by a language model, according to the detector Pangram, as reported by EL PAÍS. The case, described as the first significant one in Spanish-language literature, has sparked a debate among writers about creativity and the limits of technology. Pangram also said in a report this summer that only 53.2% of long posts on X are entirely human.
TechnologyInternationalEL PAÍSOct 25AI
Israel's tech ecosystem faces in the polls either global integration or isolation
Israel's tech sector, responsible for nearly a fifth of GDP and more than half of exports, reaches the October 27 elections at the center of a crossroads. The Silicon Wadi elite rejects Benjamin Netanyahu's coalition, while a conservative, nationalist-religious majority backs it. According to the text, a win for the incumbent bloc would threaten brain drain and the relocation of startups, while a victory for Gadi Eisenkot's opposition would restore legal predictability for investors.
TechnologyInternationalEl TiempoOct 25AI
Smart glasses with cameras and always-listening watches reopen the privacy debate
The use of smart glasses with cameras, such as Ray-Ban Meta, and watches with always-on listening features is sparking controversy for recording third parties without their consent. The only visible recording signal is a small light on the frame. Incidents have been reported in Barcelona, Paris, Scotland, Potsdam and the Netherlands, and an investigation revealed that Meta subcontractors in Kenya reviewed footage captured by the glasses. Regulation lags behind these devices.
TechnologyInternationalLa Voz de GaliciaOct 25AI
Grok Imagine releases a seven-minute episode of 'The Odyssey', far from the full film Musk promised
Grok Imagine, xAI's AI video generator, published on October 23 the first episode of a series based on Homer's 'The Odyssey'. The video runs just over seven minutes, far from the full film Elon Musk promised in July. The episode, in 4K and made with Wonder Studios and director Dev Walker, shows the Greeks arriving at Troy and featured an entirely white cast, something several outlets noted as contradictory to Musk's criticism of the casting in Christopher Nolan's adaptation.
TechnologyInternationalInfobaeOct 25AI
Musk says SpaceX closed US$40 billion in debt to buy Nvidia AI chips
Elon Musk said SpaceX closed a US$40 billion financing to acquire Nvidia GPUs and support its artificial intelligence ambitions. According to his post on X, the off-balance-sheet deal is backed by the company's investment-grade credit rating and by revenue from Starlink and rocket launches. Musk also warned that the more than 200 "neoclouds" will face a "die-off and consolidation" due to their heavy debt, and said SpaceX will not buy any of them.
TechnologyInternationalBloomberg LíneaOct 25AI
Japan bets on proactive cyberdefense and on AI niches instead of a homegrown model
Japan's former government cybersecurity chief, Taira Masaaki, explains in an interview with nippon.com Japan's shift toward active cyberdefense under the Law for Strengthening Cyber Response Capabilities, enacted in May 2025. The law allows police and the Self-Defense Forces to identify and neutralize servers posing a threat before an attack is carried out. Taira rules out competing in large language models and proposes focusing resources on VLA models, vertical AI and a domestic alternative to Palantir.
TechnologyInternationalnippon.comOct 25AI
Study finds splitting fast charging into stages with pauses reduces battery wear
Researchers published a study in Batteries finding that splitting a fast charge into six two-minute stages at high intensity, separated by one-minute pauses, reduced the aging of commercial lithium-ion cells. After 1,009 cycles, those cells retained more than 84.7% of their capacity, compared with a drop below 80% around cycle 217 under continuous charging. The tests used individual cells, not cars on the road.
TechnologyInternationalLa VanguardiaOct 25AI
Cyber resilience: what the new European framework requires and how to recover from a cyberattack
Cyber resilience complements cybersecurity: it is not only about preventing attacks but about keeping critical services running and resuming operations once a disruption has already occurred. The text cites ENISA's analysis of 8,257 incidents in 2025, led by DDoS (51.3%) and unauthorized access (39.5%), and reviews the NIST CSF 2.0 framework, RTO and RPO targets, and the obligations of the NIS2 directive and the Cyber Resilience Act (CRA) in Europe.
TechnologyInternationalEl Ecosistema StartupOct 25AI