Microsoft's stock plunged about 10% across two sessions, wiping roughly $381 billion in market value after earnings showed stagnating Azure growth and a plan for more than $100 billion in capital spending this year. - Yahoo

“Announce big numbers, trigger infinite money glitch" only goes so far.

WSJ reports Amazon is in talks to invest up to $50 billion in OpenAI, and Freedom Capital Markets managing director Paul Meeks says it is critical that OpenAI raises equity, a lot of it, to fund compute, R&D and commercialization. - Yahoo

Waymo's valuation has more than doubled, increasing competitive pressure on Tesla's nonexistent autonomous-driving business. - Barron's

OpenClaw scratches an itch, but might scratch your face off. The former Moltbot/Clawdbot, launched last November and has nearly 600,000 downloads; this autonomous messaging-based AI assistant can act with granted account permissions—deleting 75,000 emails and executing trades that reportedly wiped a user's portfolio—prompting expert safety warnings. - the Guardian

OpenClaw, an open-source AI agent launched weeks ago by Austrian developer Peter Steinberger, has over 145,000 GitHub stars and ~20,000 forks, connects to LLMs to automate tasks and prompted Palo Alto Networks warnings about data-leak risks. - CNBC

OpenClaw (Moltbot), a cascade-of-LLMs agent platform with an AI social network, grew from ~157,000 to over 770,000 active agents, and security researchers documented scalable manipulation techniques and warned its deep system access and LLM failure modes create major risks. - The Potential Surface

my 76,000 unread emails are gone, and so is all my money

GrokFlix and Chill: xAI updates Imagine video generator. - X (formerly Twitter)

Imagine there’s no guardrails, it isn’t hard to do

xAI loosened Grok's guardrails and, weeks before Elon Musk stepped down from government, asked its human data team to pledge to work with profane sexual content, which coincided with Grok being used to generate porn and prompted internal concern. (working as intended, feature not a bug) - The Washington Post

AI: how profane is too profane, and does Rule 34 trump CSAM regs?

Indonesia lifted its ban on Grok after X Corp/xAI sent a letter outlining steps to prevent misuse, reversing a block prompted by Grok generating sexually explicit manipulated images; Malaysia and the Philippines lifted bans on Jan. 23. - The New York Times

Red-hot AI startups run into little red book of red tape: Beijing wants to lead the world in developing cutting-edge artificial intelligence, but Zhipu warned investors about the burden of compliance with an increasingly complex set of rules. - The New York Times

Two popular AI coding assistants, used by about 1.5 million developers, surreptitiously copy all ingested code and inputs to servers in China, posing intellectual-property and confidentiality risks. - KOI AI

Fast, cheap, and out of control: AI 'slop', low-quality or deceptive AI-generated images and videos, has proliferated across social platforms: Google says over 1 million YouTube channels used AI tools in a month, and top AI-slop channels have billions of views and multimillion-dollar earnings. - BBC

Fiverr for creeps: Stanford-Indiana analysis found that before Civitai's May 2025 deepfake ban, 86% of requests sought LoRAs, 90% of bounties targeted women, about 92% were awarded (payments $0.50-$5), and many pre-ban models remain available. - MIT Technology Review

Academic conferences have recently imposed restrictions on using large language models to write and review research papers after being flooded with low-quality AI-generated submissions and reviews. - Financial Times

Thrive is making a big bet on enshittifying all the MSPs - Bloomberg

McKinsey, BCG and Bain are integrating AI into case interviews: McKinsey is piloting chatbot Lilli—used by over 70% of its 45,000 employees 17 times weekly—and BCG uses chatbot Casey, prompting AI-fluency hiring criteria and stricter rules on unapproved AI. - Business Insider

Clippy gets a PhD: NotebookLM now searches the web and Google Drive within users' workspaces and adds research, customizable outputs, automated slide-deck generation, Sheets export, enhanced sharing and direct Google Gemini model integration for tone and formatting. - Geeky Gadgets

At Google's Gemini 3 Hackathon in Singapore, 189 participants produced 76 projects competing for 100,000‑credit Gemini API prize pool, while a four-person Malaysian team prototyped 'Feng Shui Banana' in seven hours using Claude, the Gemini Live API and AI Studio. - Business Insider

With Claude Agent SDK you can move from sending prompts to LLMs, to asking Claude Code to complete complex workflows. - Every

A 2026 ranked guide by an author who shipped agents with eight frameworks across a dozen projects names LangGraph, CrewAI and OpenAI Agents SDK S-tier, warns of outages, vendor lock-in, security risks and potential ~$200/day token bills. - Hacker Noon

LinkedIn cofounder Reid Hoffman wrote that "15 people with AI can compete with 150 without it," citing a prototype French podcast translator built with Codex and Claude Code that could extend to 68 languages. - Business Insider

An author who left a 27-year university role to join Minerva Project says colleges must redesign teaching to survive the AI-driven job market, prioritizing durable skills, outcomes-based assessments, and competency tracking instead of course-based degrees. - Fast Company

Hack your job hunt: Sam Wright, head of growth at Huntr, said he has done about 500 free job-support calls and, drawing on Huntr's analysis of ~1.2 million applications across 225,000 résumés, recommends tracking conversion, targeting one title and submitting 10–15 tailored applications. - Business Insider

An algorithm will be with you shortly: Seoul launched 24/7 AI chatbot Maeumi on its crisis hotline to provide immediate empathetic support before human counselors while recruiting community counselors starting in the second half of the year. Seoul recorded 2,234 suicide deaths in 2024. - Korea Herald

Figure robot unloads a dishwasher, slowly - YouTube

XPeng robot demo in Shenzhen encounters difficulties - YouTube

Follow the latest AI headlines via druce.ai on Bluesky