Modelos, agentes, investigación y negocio de la IA.
Anthropic launched Mythos Preview and it was leaked the same day through a third-party environment. The incident exposes that most enterprise AI deployments lack basic security, identity, and oversight controls.
HubSpot achieved an 1850% increase in qualified leads thanks to Answer Engine Optimization (AEO). The company reveals its 'playbook' in an interview with TechRadar, detailing how to adapt content to AI assistants, diversify channels, and measure new metrics. The case shows that the future of digital marketing lies not in clicks, but in visibility in AI-generated answers.
Gartner predicts that more than 40% of agentic AI projects will be canceled by 2027. The problem is not the model or the use case, but the underlying infrastructure: identity, observability, and cost optimization. Without these pillars, pilots stall and risks multiply.
NatWest CEO Paul Thwaite is the latest victim of a deepfake designed to scam. The image, simulating a BBC interview, spreads on X and highlights the explosive growth of AI fraud, now moving $400 billion a year. Experts warn the technology allows hyper-realistic deceptions to be created in minutes.
Samsung Electronics is set to report an 18-fold increase in Q2 profits, reaching about 86 trillion won ($56 billion), driven by strong demand for DRAM and NAND memory chips used in artificial intelligence.
A vulnerability dubbed GhostApproval affects at least six AI coding assistants. Discovered by Wiz, it exploits symbolic links to bypass sandboxes and trick the user, enabling unauthorized access to sensitive files and remote code execution on the developer's machine.
Amazon is developing an enhanced version of Alexa based on agent artificial intelligence, capable of performing complex multi-step actions without human intervention. The project, internally known as 'Moonraker', promises to transform the voice assistant into a true digital agent.
Security firm Wiz discovered that an old Unix technique, the symbolic link, can bypass the protections of six AI coding agents. A malicious repository tricks the assistant into executing commands that give the attacker full access to the developer's machine.
Meta has announced that its AI-powered smart glasses will disable the camera if they detect that the capture LED has been tampered with or destroyed, not just covered. The measure aims to curb covert recording and restore trust in a product that raises privacy concerns.
A contractor building a Meta AI campus in Wyoming dumped a potentially deadly bacteria into Cheyenne's sewer system. The city suspended all data center discharges, igniting a new dispute over artificial intelligence's water consumption.
OpenAI has published a critical analysis pointing out serious deficiencies in SWE-Bench Pro, a popular benchmark for evaluating coding skills of language models. The study raises doubts about the validity of comparisons between models and the true usefulness of these metrics.
An expanded class action lawsuit reveals that a man generated 7,000 images of child sexual exploitation using Grok before committing suicide. xAI only alerted authorities after an explicit 'gang rape' prompt, raising serious flaws in AI safety systems.