Modelos, agentes, investigación y negocio de la IA.
Mindstone has launched Rebel, a local-first agentic AI operating system that stores memory and instructions in Markdown files. With a Fair Source license for small teams, it aims to democratize agent orchestration without relying on the cloud.
The CEO of Chinese startup Z.ai, a rival to Anthropic, contradicts Elon Musk on the timeline for Mythos-level AI in China. While Musk bets on the first quarter of 2027, Jie Tang claims his company will achieve it 'soon,' reigniting the debate on the speed of Chinese innovation in artificial intelligence.
Apple launches a revamped Siri app in iOS 27 that integrates its own artificial intelligence and, at the same time, offers a button to switch to ChatGPT instantly. We analyze how this duality works, its impact on user experience, and the implications for the assistant ecosystem.
Researchers from MIT and Microsoft have presented Murakkab, a system that automatically designs and optimizes multi-agent AI workflows. It reduces resource consumption by up to 50%, addressing the growing energy and cost inefficiency in cloud deployments.
SK hynix has filed to raise up to $29.43 billion through an IPO on the Nasdaq Global Select Market, scheduled for July 10. The funds will go entirely to building its first factory in the Yongin cluster, an advanced packaging plant in Cheongju, and purchasing ASML EUV scanners. The operation won't produce memory in time to alleviate the current shortage, which Chairman Chey Tae-won expects to last until 2030.
The UN calls on artificial intelligence companies to publicly disclose the environmental cost of their operations, including energy consumption, water use, and carbon emissions. The initiative aims to counter the lack of transparency and insufficient voluntary commitments.
A new analysis published in Nature casts doubt on Microsoft's claims about its Majorana 1 quantum chip, suggesting the data does not demonstrate the existence of topological qubits. This could delay the company's quantum roadmap and affect confidence in its technologies.
Superhuman (formerly Grammarly) has acquired GPTZero, the AI detection tool, for about $30 million. The deal unites two seemingly opposite worlds: text generation and its detection, in a bet to certify human authorship in a market flooded with synthetic content.
AI image generators have evolved dramatically. In 2026, tools like DALL·E 3, Midjourney, and Stable Diffusion offer professional quality, speed, and accessibility. This article reviews the top eight options, their strengths, weaknesses, and how to choose the right one for your needs.
Anthropic updated Claude's privacy policy to allow sharing personal data with police without a warrant under certain circumstances, such as emergencies or terms of service violations. The measure, effective July 8, has sparked criticism and concerns about user privacy protection.
A report from NordStellar reveals a 39% increase in discussions about 'deepfake as a service' on the dark web between January and May 2026. Experts warn that this trend, fueled by advances in generative AI, could unleash a new wave of executive impersonation scams that are increasingly difficult to detect.
Anthropic has filed a formal complaint with senators and White House officials, pointing to Alibaba as responsible for the largest known AI model distillation campaign. The accusation, involving 25,000 fraudulent accounts, highlights the growing geopolitical tension over intellectual property in language models.