Modelos, agentes, investigación y negocio de la IA.
During a shortened holiday week, US startups announced massive funding rounds, led by Joulent ($1.75B) and Together AI ($800M). AI, energy, and biotech dominated transactions, signaling a shift in investment priorities toward critical infrastructure and frontier technologies.
Anthropic and Micron have announced a four-point agreement integrating Claude models into the design and operation of memory and storage for AI. The collaboration promises to optimize HBM, DRAM, and SSD performance but leaves out computational storage, a key emerging technology.
Cloudflare has announced a new tool that allows websites to filter trackers used by artificial intelligence companies to collect data. The one-click measure gives site owners unprecedented control over how their content is used in training AI models, amid growing tension over unauthorized scraping.
Alibaba researchers present SkillWeaver, a framework that introduces Skill-Aware Decomposition (SAD) to reduce token consumption in AI agents by over 99%. The technique routes complex tasks to the right tools without loading the entire library, improving accuracy and efficiency.
Google and Amazon, two of the biggest drivers of artificial intelligence, face a dilemma: their climate ambitions clash with AI's voracious energy appetite. Their emissions have increased instead of decreasing, and net-zero promises for 2030 seem increasingly distant.
The Bank for International Settlements (BIS) warns in its 2026 annual report that massive investment in artificial intelligence, led by tech giants, could create a bubble whose burst would trigger a global recession. It compares the phenomenon to the canal, railway, electrification, and dot-com bubbles.
Cloudflare has announced that starting in September, it will block AI crawlers that do not pay publishers for content used to train models. The move could redefine the balance of power between AI and content creators, forcing companies like OpenAI and Google to negotiate licenses.
Tenet Security researchers demonstrated an 'agentjacking' attack that, using a public Sentry credential, gets agents like Claude Code to execute malicious code with developer privileges. 85% of tested agents were vulnerable, and over 2,300 organizations expose credentials that would enable the attack at scale.
Global startup investment reached $510 billion in the first half of 2026, driven by artificial intelligence. OpenAI and Anthropic account for 43% of the capital, while IPOs and acquisitions hit a historic record.
An article from Towards Data Science warns about 'prompt regression': seemingly innocuous modifications to language model instructions can cause silent failures in production. This analysis explains the phenomenon, its impact, and how to mitigate it.
Agentic artificial intelligence promises to transform business automation by enabling systems to make decisions and execute tasks independently. A new report reveals that 84% of business leaders plan to increase their investments in this technology over the next year. We analyze what agentic AI really is, why it is gaining traction, and what implications it has for businesses.
Three weeks after the Trump administration classified Claude Fable 5 and Mythos 5 models as national security risks, the U.S. government has lifted export restrictions, allowing global deployment. Anthropic worked closely with authorities to implement safeguards, paving the way for a new era of regulated frontier AI.