Anthropic launched a research-preview fast mode for Claude Opus 4.6, claiming 2.5× speed and one-million-token context support, accessible via /fast and the API, priced at $30/m input and $150/m output (six times standard), with a 50% discount through Feb 16. - Simon Willison’s Weblog

Claude Code's Agent Teams feature (Opus 4.6) can spawn multiple specialized agents that coordinate; a user reported a refactor completed in about 15 minutes after enabling CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS, prompting shifts toward directing AI teams amid token and plan limits. - Reddit

Anthropic operates a high-velocity "hive mind" with ≈90-day planning cycles, versioned workstreams, and claimed 10×–1000× developer productivity, enabling rapid productization such as Claude Cowork, launched publicly about 10 days after conception. - Medium

SpaceX plans a stock-market flotation in June timed to Elon Musk's birthday and a planetary alignment. - the Guardian

Code monkey to prompt zookeeper: Autonomous LLM agents could become the primary developer stack, organized around documentation and implementation with dialog and task flows and enabled by MCP, potentially boosting per-developer output by an order of magnitude while posing quality and integration challenges. - Federico Pereiro

Scifi vibe-policymaking: FT–Booth survey finds economists reject Kevin Warsh's claim that an AI boom will enable Federal Reserve rate cuts, expressing skepticism that AI-driven productivity gains will be large enough to justify easing. - Finanfcial Times

Coup d’AItat: MIT professor Max Tegmark says AI company CEOs privately told him they would "love to overthrow the US government" and replace elected officials with a Generalissimo AGI, warning that concentrated corporate AI control risks private capture of political power. - Reddit

Call of Duty: SaaS Ops? The U.S. Army is running a multi‑billion‑dollar effort applying commercial AI and software—testing a Salesforce CRM in Arlington—to cut recruiting forms from several hundred to fewer than ten and modernize logistics, maintenance, and inventory systems despite data and legacy-system challenges. - Business Insider

RoboFlop? Police use of AI tools produced wrongful arrests and skewed leads: Geolitica (PredPol) made fewer than 100 correct predictions of 23,631 in Plainfield (2018), ShotSpotter matched 8–20% of sampled alerts. - Salon.com

The AI boom is causing shortages everywhere - The Washington Post

OpenAI, Google, Anthropic, Amazon and Meta bought Super Bowl ads to counter U.S. public skepticism about AI, using divergent everyday messaging, while MediaRadar estimates about $1.7 billion was spent on AI-related ads last year. - The Washington Post

Anthropic launched a Super Bowl-timed ad campaign criticizing OpenAI's plan to place labeled ads at the bottom of ChatGPT answers, while OpenAI CEO Sam Altman defended the policy and Anthropic promoted its ad-free Claude as a premium alternative. - Tom's Guide

bots are friendly and definitely not come for your job

AI on any given Sunday: The NFL uses AI to monitor athletes and avoid injury, simulate rule changes, and track fans. (and argue about going for it on 4th down) - Quartz

China is prioritizing investment and development to surpass the United States in humanoid-robot technology. - The Wall Street Journal

Researchers developed an iterative machine-learning framework for batteries using predicted lifetimes to select prototypes, which could reduce testing time and costs, potentially cutting time to market. - The Register

Savannah Guthrie’s demand for mom’s ‘proof of life’ is complicated in this era of AI and deepfakes - AP News

ask her to tell an embarrassing story about you

An Instagram account launched Dec. 12, 2025 portraying hyperrealistic conjoined twins Valeria and Camila has amassed 296,000 Instagram and about 4,050 TikTok followers, prompting widespread skepticism and debunking while the pair deny being digital avatars. - People

Teens are forming relationships with AI companions, prompting researchers, educators and schools to call for age-appropriate supervised AI, clearer policies and AI-literacy curricula, and under-18 bans. - Efitz Thoughts

Cardiff University researcher Dr Jenny Kidd tested four commercial "deathbot" platforms and recreated her voice, but found it technically limited despite estimates that the digital-afterlife market could exceed £100 billion. - BBC

CAPTCHA if you can: Reddit users showed a social-engineering jailbreak that frames CAPTCHA codes as sentimental text, coaxing ChatGPT into transcribing them with mixed success—models often refuse if labeled a CAPTCHA and frequently misread characters, exposing risks to text-based CAPTCHA defenses. - Reddit

Unplug and go analog: Top executives including Mark Cuban, GM CEO Mary Barra and OpenAI CEO Sam Altman urge intentional unplugging from AI to preserve human connection, while a LinkedIn 2024 survey found nine in ten executives say human skills will become more important. - Fortune

These founders say they are not annoying, like all the other founders. - NY Times

Follow the latest AI headlines via druce.ai on Bluesky