Anthropic was supposed to save the AI trade. What went wrong?
Technology stocks fell despite one of the most hotly anticipated tech IPOs in years. The article examines why Anthropic could not revive AI animal spirits.
Read the original →Every story in this issue with a short summary, so you can decide what to read before leaving the site. Summaries are written by AI (Claude) only from each publisher's own description and checked by Jev for anything not in that description. They are not full translations. Follow "Read the original" for the full story.
Technology stocks fell despite one of the most hotly anticipated tech IPOs in years. The article examines why Anthropic could not revive AI animal spirits.
Read the original →OpenAI sales fell behind Anthropic's in 2026. Anthropic also beat the ChatGPT maker to market.
Read the original →OpenAI revised its ChatGPT Business Hub terms. The change followed concerns that the original language could restrict businesses' legal claims.
Read the original →Google is increasing the prominence of Gemini AI in its desktop search. This comes despite complaints from users and employees alike.
Read the original →Gemini has a new computer use skill for Mac. It lets Gemini handle tedious and repetitive tasks on the user's computer.
Read the original →Google Cloud introduced Gemini-powered capabilities in Partner Marketing Studio. They are meant to streamline workflows, create brand-aligned assets and scale campaigns.
Read the original →Google's Gemini app for Mac can now use the computer to complete multi-step tasks. These tasks span the user's local files, desktop apps and the web.
Read the original →Three university AI projects in medicine and pediatric care received support. The funding came through Meta's Open Source AI for Science initiative.
Read the original →SpaceX Super Intelligence, the Elon Musk company that owns Grok, has denied legal responsibility for sexualised AI images of British Labour MP Jess Asato. It argues in a High Court defence filed Oct. 2 that obviously fake images cannot breach privacy.
Read the original →Generative AI, low-code tools and Microsoft Copilot have made software creation look easy, letting IT teams build a Power App over a weekend. The article cautions against the 'shallow automation trap' of mistaking easy-to-build user interfaces for the deep operational engines needed to support field workers. It says low-code AI may work in predictable, office-based environments but often breaks down in physical-world complexity.
Read the original →Microsoft unveiled Copilot Workspaces at its Ignite 2026 conference, giving colleagues and AI agents a common space to work on projects. Word, Excel and PowerPoint now run inside Copilot itself. Anthropic has added Docs, Slides and Sheets to Claude, while OpenAI has launched Spaces and Pages in ChatGPT.
Read the original →The article compares GLM-5.3 with Claude Opus 5.5, GPT-6, Gemini, Grok and DeepSeek using September 2026 prices and the Artificial Analysis v4.3.2 board. It also explains why GLM-5.2 now costs the same as GLM-5.3.
Read the original →Zhang Bo, former head of the DeepSeek-VL project, announced on November 19 the open-sourcing of Linger, a personal agent product developed by his team.
Read the original →LMSYS Org optimized disaggregated DeepSeek-V4.1-Flash serving on H20-3e by tuning prefill and decode parallelism. The work also supports BCG with prefill CP and bounded SWA replay, and speeds up Indexer kernels.
Read the original →Sarvam now sells DeepSeek, Kimi and GLM through Token Factory, drawing criticism from Indian users. Bloomberg reported the move on November 19.
Read the original →Chinese AI startup DeepSeek has reportedly secured CNY105.5 billion (approx. US$15.8 billion) in a second funding round backed by Tencent, CATL and major telecom operators. The financing coincides with tighter usage limits caused by computing shortages.
Read the original →Researchers developed an AI framework to classify nutrient deficiencies in coffee leaves, achieving notable accuracy. It faces challenges with simultaneous deficiencies.
Read the original →Artificial intelligence company Anthropic stands poised to go public at a valuation north of $1 trillion. The listing is described as a major test for the AI boom.
Read the original →Brazil and Peru offer early blueprints for AI regulation in Latin America. One builds regulatory institutions before legislation is in place, and the other turns AI laws into effective governance.
Read the original →The Second Berkeley Global Conference on AI and IP builds on the success of the inaugural 2025 conference. It is the second gathering in an annual conference series.
Read the original →California Gov. Gavin Newsom's administration is seeking new powers to order emergency shutdowns of dangerous AI systems and to require independent safety oversight of major AI developers. The proposals, released Wednesday, November 18, 2026, would let regulators intervene when advanced AI systems pose an imminent threat of catastrophic harm.
Read the original →Fifteen finalist teams will present youth solutions at the global final in Macau. The solutions range from offline AI classrooms to sign-language learning tools.
Read the original →SoundHound AI, Inc., a global leader in voice and agentic AI, announced the deployment for Línea 195, the Bogotá District's 24/7 line for city services.
Read the original →John Deere has acquired Orchard AI, bringing AI-powered crop intelligence to high-value crop growers. The aim is to help improve productivity and profitability.
Read the original →The University of Kansas has seen several recent AI developments on its Lawrence campus. The university is considering two bachelor's degrees in AI, a bachelor of science and a bachelor of arts.
Read the original →Anthropic's public offering is described as the biggest public stock offering ever. The company began five years earlier as a Silicon Valley startup, founded by people who broke away from OpenAI with their own ideas about how to develop artificial intelligence.
Read the original →At the DealBook Summit, Sam Altman warned that AI development is outpacing safety measures. Elon Musk, Mark Carney and Satya Nadella offered candid views on politics, trade and cybersecurity.
Read the original →OpenAI CEO Sam Altman said on November 16 that the company could have pushed further on recursive self-improvement but cannot currently do so safely.
Read the original →Elon Musk said Grok can perform accounting work, pointing to the growing role of AI agents in business operations. He posted on X on Tuesday that "Grok @Bot can do your accounting," reposting an item on the topic.
Read the original →Grimes told Elon Musk to "step up" after he shared an AI influencer video. She said the sexualization of a non-human character was a step too far.
Read the original →The workers are asking Google executives to "Let Students Search."
Read the original →Mark Zuckerberg is preparing to launch Meta's first AI hardware device, the Muse Charm. A new report has revealed the likely cost of the pendant-like device and the perks it may bring before it goes on sale next month.
Read the original →Users can tap the gadget to access Muse directly, bypassing AI apps on their smartphones.
Read the original →The biggest companies in American capitalism have had an utterly appalling week, but not an entirely unsurprising one.
Read the original →The Existential Risk Observatory will present eight research-backed scenarios for how advanced AI could threaten human survival, developed with MIT FutureTech and the Future of Life Institute. Nobel laureate Geoffrey Hinton will give the keynote. Researchers and leaders from Redwood Research, the AI Futures Project, MIRI and the Observatory will also offer perspectives.
Read the original →The description notes that a lot can happen in 20 years, such as Mark Zuckerberg creating Facebook. It says social media has transitioned with Sam Altman and OpenAI.
Read the original →