Nachrichten — 2026-08-13

Anthropic announced that its Claude models will embed an imperceptible watermark in every piece of generated text, a move designed to satisfy the European Union's AI Act requirements for machine‑readable provenance. The watermark, applied at the model level, survives copy‑and‑paste and minor edits, and will be retroactively added to existing Claude deployments. The change comes as the EU mandates detectable AI‑generated content by August 2, 2026, with a transition period lasting until December 2, 2026. Industry observers wonder how the measure will affect Claude users and whether rivals OpenAI and Google will adopt similar tactics.Weiterlesen

Google announced Tuesday that its Gemini AI app has surpassed one billion monthly active users, making it the company’s 14th product to reach that scale. The milestone arrives ten weeks after OpenAI’s ChatGPT hit the same mark, narrowing the gap between the two AI assistants. Google highlighted rapid growth, strong iOS adoption and expanding voice and image capabilities, while noting the product is still evolving and that weekly‑active figures remain undisclosed.Weiterlesen

Thrive Holdings announced a $2 billion funding round that values the AI‑focused private‑equity firm at $12 billion. Backers include SoftBank, D1 Capital Partners and Altimeter Capital. The capital will fund the firm’s next growth phase, adding a regulatory‑services platform for physical‑asset projects and deepening its partnership with OpenAI. Thrive, a spin‑out of Thrive Capital, already runs AI‑enhanced accounting and IT businesses, reporting dramatic productivity gains and a growing roster of more than 70 companies on its platforms.Weiterlesen

Cognition, the maker of the AI coding agent Devin, is reportedly in early talks with investors for a fresh financing round that could push its valuation to $40 billion. The move follows a $1 billion raise at a $26 billion valuation in May and hinges on the company hitting a $1 billion annualized revenue run rate, according to Bloomberg‑cited sources.Weiterlesen

Nvidia unveiled Nemotron 3.5 Lightning, a 30‑billion‑parameter mixture‑of‑experts model that runs on a one‑million‑token context window and is free for commercial use under the OpenMDW‑1.1 license. The company also released the NeMo Switchyard library, which routes tasks to the most suitable model, cutting costs by up to 74 percent. While Nemotron 3.5 outpaces comparable models in raw speed, Nvidia’s longer‑term goal is a trillion‑parameter Nemotron 4 that could still lag behind China’s largest open models such as Moonshot’s Kimi K3.Weiterlesen