Modelos, agentes, investigación y negocio de la IA.
Mamba, a model based on state-space models (SSMs), challenges the dominance of Transformers by offering comparable performance with much higher efficiency on long sequences. Its linear architecture eliminates the quadratic bottleneck of attention, enabling contexts of up to a million tokens and speeds up to 5 times faster.
At the upcoming VivaTech 2026, enterprise artificial intelligence will take center stage, reflecting a European trend: integrating AI into existing complex systems rather than pursuing massive consumer language models. This approach promises efficiency and sustainability but poses regulatory and scalability challenges.
Dario Amodei, CEO of Anthropic, has published an essay calling for government regulation of frontier AI models, inspired by the FAA. The proposal includes mandatory third-party testing, compute thresholds, and the ability to block deployments that pose a risk to public safety. This would have direct implications for companies integrating AI into their operations.
For the first time, a Western government applies real-time export control on a commercial AI already deployed. The order forced Anthropic to globally disconnect its most advanced models, Fable 5 and Mythos 5, after an alleged jailbreak that would have allowed access to offensive cybersecurity capabilities. The precedent marks a before and after in AI regulation.
OpenAI has announced a ChatGPT memory update that allows the assistant to remember preferences, tastes, and relevant information between sessions. This change promises more coherent and personalized conversations, but also raises questions about privacy and data control.
Code agents are transforming the daily work of software engineers, but not always for the better. A recent article from the Stack Overflow Blog points out that decision fatigue has become a critical side effect: every suggestion, every generated code block, and every review demands constant decisions. This analysis explores the causes, consequences, and possible solutions to a problem that could slow down AI adoption in development.
NVIDIA announces the global expansion of its AI Cloud ecosystem, with new partners in Africa, South America, and other regions. The move aims to bring AI infrastructure closer to local data centers, reducing latency and supporting digital sovereignty.
DeepSeek has launched V4, a model with a one million token context capacity, optimized for AI agents to process extensive documents, complete codebases, and long conversations without losing accuracy or performance. This breakthrough promises to transform enterprise and development applications.
OpenAI has identified an influence operation attributed to China that used its own chatbot to create social media posts. The campaign sought to fuel opposition to AI data centers and criticize Trump's tariffs, though it had little real impact.
Coinbase has introduced Coinbase for Agents, a tool that enables AI agents like ChatGPT or Claude to execute trading operations and manage cryptocurrency payments autonomously. Using the x402 protocol, these agents can pay for digital services without human intervention, a step toward an economy where AI agents act as new financial players.
Visa and OpenAI announce a collaboration to manage transactions initiated by artificial intelligence. While promising advanced security, questions arise about consumer trust and control in an increasingly autonomous ecosystem.
A multidisciplinary team has developed Latent Context Language Models (LCLM), a new architecture that compresses input context up to 16x without significant loss of accuracy. This solves the computational bottleneck of context windows in LLMs, especially in autonomous agents.