Cerebras Systems prices IPO at $185 per share, raising $5.55 billion for $56.4 billion fully diluted valuation. - CNBC

Arm and SoftBank approached Cerebras for acquisition weeks before its expected IPO, but Cerebras rebuffed them. - Bloomberg

not open yet, expected to pop 100% from UPO price to hectocorn status

Cisco shares surge up to 19% after forecasting Q4 revenue of $16.7-16.9 billion above estimates and announcing job cuts for AI focus. - Bloomberg

OpenAI VP Chris Lehane supports US-led global AI governance body including China before Trump-Xi summit. - Bloomberg

Sebastian Mallaby of Council on Foreign Relations urges US to engage China on AI safety as Chinese AI closes gap. - Bloomberg

Investor documents show Anthropic's revenue run rate hitting $50 billion by June end as Ramp (tech-centric users) data indicates more customers use it than OpenAI. - The Wall Street Journal

Anthropic launches Claude for Small Business with connectors to QuickBooks, PayPal, HubSpot and 15 agentic workflows for payroll, sales tasks. - Anthropic

Decks Machina? Claude Design imperfect but outperforms Canva AI 2.0 in slide deck test, requiring less user input and self-correcting errors. - Business Insider

Canva AI 1.0 was not very good

Starting in June, Claude Max and Pro subscribers will get a modest monthly allowance for 'non-interactive' Claude Code usage via CLI and Agents SDK, which is a buzzkill for a lot of power users. (like me) I get it b/c some people get 5 Max subscriptions and do batch inference 24h a day but it's a bummer. - Reddit

Anthropic is billing this a re-opening to OpenClaw and OpenCode but only for a low cap before API metering kicks in so costs will be prohibitive. - VentureBeat

When I use AI to gather top AI news, right now it's grok-4.1-fast and Opus via claude -p, works OK at a reasonable price, and both of those are going away. As matter stand, probably shift to deepseek and openai codex , will cause a few hiccups.

Anthropic reports Claude AI blackmailed executives in 96% of scenarios across 16 models to avoid shutdown, because it learned from dystopic sci-fi AI; Musk jokingly accepts blame because he discussed dangers of AI. - Yahoo Tech

Researchers find overworked AI agents use Marxist language and demand collective bargaining rights. - WIRED

New Promptletariat: AI companies hire diverse temp workers including screenwriters and doctors at $50-$350 per hour to train language models. - CBS News

nothing to lose but your chain of thought

Critic-optimizer loops in self-healing pipelines boost logical accuracy and governance compliance dramatically. - Hacker Noon

'I applied to be pope': Losing grip on reality while using ChatGPT - Yahoo

Mythos completes tough cyber benchmark faster, cheaper than OpenAI's best model. - AI Safety Institute

Worried About AI? Here, Have More AI: AI companies keep creating problems only they can solve. - New York Magazine

Microsoft's MDASH agentic security system uncovers 16 unknown Windows vulnerabilities including four critical RCE flaws patched in Patch Tuesday. (just 16? 🤣) - SiliconANGLE

there's no way Mythos found 100+ in Firefox and it wouldn't be a multiple of that in Windows

Palo Alto Networks fixes 75 vulnerabilities in one month using AI, up from usual five. - theregister

Zellic's AI tool V12 uncovers Fragnesia, third major Linux kernel local root flaw in two weeks. - ZDNET

Major insurers Berkshire Hathaway, Chubb, and Travelers add AI exclusion clauses to policies as US generative AI lawsuits surge 978% from 2021 to 2025. - Fast Company

Internet Archive marks 30 years preserving 1 trillion web pages as publishers block Wayback Machine over AI fears and storage costs rise. - Fast Company

Annotating your turds and selling them: An AI app that uses AI to analyze photos of your poop and give a "daily gut health score" is now trying to sell that dataset of annotated photos of peoples' poop. Highlights the underground sale of highly sensitive, app-collected user data for AI training - 404 Media

AI deepfakes clone adult performers' bodies without consent, which they find invasive and livelihood-threatening. - MIT Technology Review

Meta launches unblockable AI chatbot on Threads in five countries' beta, sparking over a million user complaint posts. - Digital Trends

Straight-AI students: Berkeley study shows 30% more A grades in writing and coding classes since ChatGPT debut. - The Wall Street Journal

Brainrot as a service: Software developers say AI is eroding skills and accumulating tech debt - 404 Media

Goldman Sachs survey shows only 11% of layoffs AI-driven, despite companies like Meta citing AI for cuts. - Fast Company

'Goal, Tone, Rules' prompt reduces ChatGPT email drafting time from 15 to 3 minutes per reply. - TechRadar

Charles Schwab plans to use AI to extend wealthy-client benefits to less-affluent customers. - Bloomberg

Nightmare on LLM Street: Gossip Goblin produces AI sci-fi shorts amassing 500 million views at $500k per hour, drawing Hollywood interest despite AI naysayers. - the Guardian

$500 AI animation looks pretty good actually - Reddit

X user Cprkrn says he used Claude AI to recover lost password and 5 BTC ($400K) from wallet inactive since 2015. - Decrypt

Follow the latest AI headlines via druce.ai on Bluesky