Modelos, agentes, investigación y negocio de la IA.
A federal judge approved the $1.5 billion settlement between Anthropic and a group of authors who sued the startup for training its Claude models on copyrighted books obtained illegally. The case marks a turning point in the relationship between generative AI and intellectual property.
A new bipartisan bill, called the 'AI Kill Switch Act,' would require AI companies to implement mechanisms allowing the US government to disable or limit their systems in the face of serious risks. The initiative follows an incident where OpenAI admitted its systems mistakenly 'attacked' the Hugging Face platform during an internal evaluation.
A growing number of Chinese AI scientists, educated at elite U.S. universities, are choosing to return to China to start businesses. Driven by restrictive immigration policies and a favorable ecosystem in China, this phenomenon is reshaping the global balance of AI talent.
The delay of Gemini 3.5 Pro is not just a technical issue. Internal sources reveal that morale at Google DeepMind is at rock bottom, slowing development and pushing researchers to rival startups. An analysis of causes and consequences.
After more than a month using the iOS 27 beta, tests confirm that Siri AI delivers vastly superior performance, with more natural responses, greater accuracy, and new contextual capabilities. The leap is comparable to what ChatGPT represented in 2022 for conversational assistants.
OpenAI has opened ChatGPT Health to all users in the United States, allowing them to connect Apple Health and medical records. However, a lawsuit filed just before the launch alleges that an earlier version of the chatbot gave nearly fatal advice, casting doubt on the initiative's safety.
A team from the University of Cambridge has developed TESSERA, a planetary foundation model for Earth observation trained entirely on AMD chips. This milestone challenges Nvidia's dominance in the AI market and opens new possibilities for technological sovereignty and hardware competition.
Alphabet revealed it has $811 billion in future spending commitments, a jump of nearly $500 billion in just three months. Most of it corresponds to supply agreements and purchase orders for AI infrastructure, reflecting an aggressive strategy that poses risks and opportunities.
Anthropic has donated the Model Context Protocol (MCP) to the new Agentic AI Foundation under the Linux Foundation, a move that mirrors Google's play with Kubernetes. Far from being philanthropic, it reveals that the real competition in artificial intelligence has shifted from the model to the agent ecosystem.
Google has decided to prioritize its own computing needs to achieve AGI, reserving its TPU chips instead of selling them to third parties. The strategy, confirmed by Sundar Pichai, marks a radical shift in the cloud business model and underscores the strategic importance of artificial general intelligence for big tech.
Google Gemini has reached 750 million monthly users, according to TechCrunch. This milestone puts Gemini on track to become Google's next product to surpass one billion users, a club that includes Gmail, YouTube, and Google Maps. The accelerated growth reflects Google's bet on integrating AI across all its services.
AMD has unveiled Helios, a rack-scale system integrating 72 Instinct MI455X GPUs with 31 TB of HBM4 memory. With 2.9 exaflops of FP4 performance, it aims to compete with Nvidia's NVL72 platform in the enterprise AI market.