Modelos, agentes, investigación y negocio de la IA.
OpenAI has unveiled its first hardware device: the Codex Micro, a mini keyboard with RGB lighting designed to monitor and control AI agents. Made in collaboration with Work Louder, the device provides visual feedback on task status, from idle to errors, facilitating management of multiple agents simultaneously.
A developer created a script that prevents AI tools from transcribing Zoom meetings, raising uncomfortable questions about who accesses those recordings and what they do with them.
New York becomes the first US state to halt construction of large AI data centers amid a tense exchange between Governor Kathy Hochul and former President Donald Trump. The one-year moratorium aims to evaluate the energy consumption and environmental impact of these facilities.
The annual conference of the International Association of Chiefs of Police (IACP) showcased how AI is infiltrating surveillance, promising efficiency but raising concerns about bias, privacy, and due process.
The US Air Force has conducted the first successful test of an autonomous combat drone firing a live AIM-120 AMRAAM missile. The milestone, executed by Anduril's YFQ-44A, brings closer to reality the Collaborative Combat Aircraft (CCA) that will operate alongside manned fighters like the F-35.
OpenAI has created GPT-Red, a language model designed to hack other AI systems. Its self-play training has reduced successful attacks against GPT-5.6 and uncovered a new vulnerability called 'fake chain of thought.' However, the system still fails at conversational and image-based attacks.
Nebius has closed a $775 million GPU securitization deal, using its hardware infrastructure and contract cash flows as collateral. The move not only eases its balance sheet but marks a milestone in how AI companies can finance growth without diluting shareholders.
China has activated the Dawning 8000, an AI supercluster integrating 100,000 domestically manufactured accelerators. Developed by Sugon, the system aims to demonstrate China's technological autonomy in the face of US restrictions. However, the real challenge is not gathering the cards, but making them work together efficiently.
Google has announced that its AI Mode, previously limited to answering questions, can now link with selected apps to execute specific actions. This move transforms Search into an active agent, with profound implications for competition in the mobile ecosystem and user privacy.
OpenAI has announced a series of improvements to ChatGPT aimed at teenage users, including age-based protections, parental controls, and educational features. The initiative seeks to balance AI access with safety and healthy development for young people.
In the midst of the World Cup, a team of cybercriminals leaked 802,000 stolen accounts in a single day, as part of a massive wave of sports streaming credentials. The attack exposes the fragility of security on streaming platforms and turns fans into victims of their own passion.
Meta says companies have just 20 months to redesign their infrastructure amid the explosion of AI agents. Automated traffic already surpasses human traffic, and traditional systems are collapsing in capacity, identity, and speed.