Articulos
Un feed continuo de noticias de IA, lanzamientos de productos, cambios de plataforma, movimientos de startups y otras novedades tecnologicas.
General Compute’s $400 million financing, backed by inference-focused chips rather than training GPUs, suggests AI infrastructure investors are starting to treat inference hardware as financeable collateral and a distinct growth market.
Patreon says it is escalating its anti-scraping strategy by using Cloudflare tools to actively block AI training bots, reflecting a broader shift across the web from voluntary crawler rules to enforceable controls over model-training access.
Google is adding app integrations to AI Mode, letting users connect services such as Instacart, Canva, and YouTube to complete tasks directly from its conversational search interface. The move pushes AI Mode beyond answers and into action, strengthening Google’s position against rival AI assistants that already support connected tools.
xAI has filed a federal lawsuit against a Grok user accused of trying to generate child sexual abuse material and non-consensual deepfake pornography, marking a rare case of an AI company taking direct legal action against a user over alleged criminal misuse of its model.
Anthropic has revealed how its engineering teams used Claude Fable5, Opus4.8, and dynamic workflows to migrate 10 large codebases within a month. Two standout projects included Bun's migration from Zig to Rust in under two weeks and Anthropic Labs' conversion of a 165,000-line Python codebase to TypeScript over a single weekend. The company also outlined.
Generative AI music platform Suno has suffered a supply chain attack that exposed its internal source code and data collection details, revealing that it had scraped large volumes of music, lyrics, and audio materials from multiple platforms for AI model training. The incident also involved the leak of some user database information and could further intensify the copyright litigation pressure facing Suno.
AI-powered travel platform Fora has raised $60 million in Series D funding, reaching a $1 billion valuation and officially becoming a unicorn. The company plans to use the new funding to expand the adoption of its AI assistant Via and further develop travel services including cruises and flight bookings.
Ahead of the 2026 World Artificial Intelligence Conference (WAIC), Moonshot AI has launched its next-generation large language model, Kimi K3, featuring 2.8 trillion parameters and becoming the world's largest open-source model by parameter count. The model also supports native vision understanding and a 1 million-token long context window, marking a significant breakthrough for the open-source AI community in terms of model scale and complex task capabilities.
Unisound has launched the U2-Med Tri-Medical Model, built on its upgraded U2 foundation model, positioned as the industry's first large model targeting medical services, insurance, and pharmaceuticals together. It aims to improve coordination across hospitals, insurers, and drug operations, and has already won AI project bids at several major Chinese hospitals. Unisound frames it as part of a broader push into domain-specific healthcare AI.
Samsung's Galaxy AI has officially received regulatory approval for its generative AI service in China, with Baidu AI Cloud serving as its technology partner. The service has been included in the latest batch of approved on-device generative AI services for smartphones, underscoring Samsung's continued efforts to strengthen its localized AI capabilities and ecosystem in the Chinese market.
Applied Computing, a London-based startup building a foundation AI model for the oil, gas, and petrochemical industry, has raised a $20 million Series A led by KBR with participation from Databricks Ventures. The company says its Orbital product can help operators detect anomalies, investigate causes, and model fixes across industrial facilities in minutes.
**The latest offline IQ benchmark from Tracking AI shows that multiple variants of OpenAI's GPT-5.6 achieved an IQ score of 136, marking the first time a large language model has surpassed the 130-point threshold. Beyond its benchmark performance, several developers report that GPT-5.6 also excels in real-world applications, including web development, building retrieval-augmented generation (RAG) ticketing systems, and software bug fixing.**