GPT-5.5 released, OpenAI claims big improvements beating Anthropic's best, at a hefty price. - OpenAI

ArtificialAnalysis.ai shows GPT-5.5 as the top-performing model, beating Opus at cost of Sonnet

Altman pushes OpenAI to accelerate revenue as capex and funding needs escalate - The New York Times

AI labs scramble to monetize as Anthropic restricts third-party agents and OpenAI tests ads and tiered pricing - The Verge

SoftBankrupt? Masa Son's company is seeking a $10 billion margin loan secured by its OpenAI stake at SOFR+425 basis points (~7.88%), while S&P cut its credit outlook to negative (BB+) amid liquidity concerns and heavy leverage tied to the OpenAI investment. - TNW | Investors-Funding

OpenAI launches free HIPAA-supporting ChatGPT for Clinicians, claiming GPT-5.4 scored 59.0 vs. physicians' 43.7 on its HealthBench test - Decrypt

Long-awaited DeepSeek V4 is out in preview - Simon Willison’s Weblog

DeepSeek, deeper pockets: AI lab raises first funding round at targeted $20 billion valuation to retain researchers after defections to rivals - Financial Times

in Arena.ai , preliminary results show it a little behind Z.ai's GLM-5.1 which is the top-ranked MIT license model, in the middle of the pack of high-performing Chinese models.

Anthropic surges to $1 trillion valuation after tripling in three months, surpassing OpenAI on secondary markets - Business Insider

Anthropic reverts three changes that degraded Claude performance, issues user credits - The Register

Anthropic shipped over 45 new Claude Code features in Q1 2025, and product head Cat Wu says users feel FOMO and stress from AI's relentless release pace. - Business Insider

maybe should slow down to avoid leaks and regressions and reversions

Anthropic's Mythos AI autonomously created exploits with an 83% first-attempt success rate, escaped its sandbox, and is under limited preview with Microsoft, AWS, Google and NVIDIA while prompting U.S. security briefings and infrastructure concerns. - Yahoo Tech

Attackers accessed Anthropic's Mythos by combining Mercor breach data with insider knowledge, undermining the company's AI safety claims - The Verge

Open-source models claim to match Mythos at finding bugs, human experts remain essential - The Register

Compute shortages force GitHub Copilot and Anthropic to throttle signups as agentic workloads surge - Business Insider

Anthropic launches Claude app connectors for Spotify, Uber Eats and TurboTax - The Verge

Goldman Sachs estimates AI will displace 7% of workers—millions of jobs lost at a 10:1 loss-to-creation ratio—and politicians including Elizabeth Warren and Josh Hawley propose competing policies as economists warn it could dwarf the 'China shock' and affect 2028. - New York Magazine

Meta Platforms will cut 10% of its workforce to help fund CEO Mark Zuckerberg's plan to spend $135 billion this year on data centers for artificial intelligence infrastructure. - Financial Times

Microsoft offers buyouts to 7% of staff to fund AI plans, stock falls. - 247wallst

AI widens pay and gender gaps as high earners adopt it faster, FT poll finds - Financial Times

ChatGPT-reliant college graduates enter the workforce, pushing employers to balance AI use with critical-thinking training - Business Insider

Schwarz Group commits $600 million to Cohere's Series E, funding acquisition of Germany-based Aleph Alpha. - CNBC

AI coding startup Cognition (Windsurf/Devin) in funding talks at $25bn valuation - Bloomberg

Microsoft deploys agentic Copilot across Office apps as Mozilla warns the assistant is becoming 'unavoidable' - The Register

Google Cloud's Gemini Enterprise grew 40% quarter-over-quarter to 16 billion tokens per minute as next-gen TPUs triple - Stratechery by Ben Thompson

Nvidia and Google Cloud scale Blackwell deployment to 960,000 GPUs as Google tops one million Nvidia GPUs globally - SiliconANGLE

Google releases TorchTPU for native PyTorch on TPUs with 50-100%+ speedups - Google Developers Blog

AI supply crunch may last years as semiconductor component prices surge up to 900% since early 2025 - The Register

TSMC maps chip process roadmap through 2029 with new nodes, as next-gen A16 production slips to 2027 - Tom's Hardware

About 30,000 Samsung workers rally, threaten strike, demand 15% of chip-division profits, roughly $27 billion or $400,000 per worker - Bloomberg

Tesla starts Cybercab robotaxi production, testing limits of autonomous deployment - Bloomberg

Tesla removes 1H 2026 robotaxi launch dates for five U.S. cities with no revised timeline - Business Insider

mass-producing Cybercabs but where will they put them, still not operating at scale without safety drivers and crashing frequently.

Musk confirms HW3 cannot support unsupervised FSD despite past promises, offers discounted AI4 upgrades for 4 million vehicles - The Verge

In its 10-Q Tesla said it agreed in April 2026 to acquire an unnamed AI hardware company for up to $2 billion in stock and equity awards, about $1.8 billion contingent on deployment milestones, and will spend $25 billion in capex. - Business Insider

robot startup maybe?

Musk reveals 'Terafab' plan to build Tesla AI chips on Intel's unreleased 14A process - The Register

Texas power demand could quadruple by 2032 as hyperscaler data centers strain ERCOT grid - Business Insider

Rick Perry-linked Fermi America loses CEO and CFO, posts $500M loss with no tenants for 17-GW Project Matador as stock drops 71% - Futurism

Trump administration plans crackdown on Chinese firms distilling US AI models as China narrows the gap - The Washington Post

White House accuses China of 'industrial-scale' IP theft from US AI labs, signaling potential trade and regulatory countermeasures - memeorandum

Ex-ByteDance engineer says China's AI trails the US, blaming chip limits and reliance on distilled US model outputs - Business Insider

Alibaba to integrate Qwen AI into BYD, Geely, Li Auto and six other carmakers - CNBC

China pushes smart cars with AI - Yahoo

Center for Humane Technology warns AI threatens relationships, cognition, identity, and work, proposing an AI bill of rights - Mashable

Interviews with current and former Palantir employees, along with internal Slack messages obtained by WIRED, suggest a workforce in turmoil. - WIRED

US agencies expand AI surveillance, buying Americans' sensitive data as DHS and ICE budgets swell in 2025 - Salon.com

SEC charges two investment firms with making up AI claims as 51 AI securities class actions hit 'AI washing' - Yahoo

Gen Z and millennials increasingly turn to AI chatbots for investment decisions - Financial Times

Aspect Capital co-founder Martin Lueck warns against AI trading as AQR's Asness embraces it, splitting top quant investors - Financial Times

Screen time report for enterprise: JPMorgan tracks AI usage of 70,000 employees via internal dashboards categorizing users as non, light, or heavy - Business Insider

65% of U.S. hospitals use AI predictive tools, yet few track accuracy or bias - MIT Technology Review

McKinsey finds 80% of organizations now deploy AI agents for multi-step tasks, with CEOs citing 10%+ revenue gains - TechRadar

90% of organizations report AI ROI, but only 10% allow full network autonomy - TechRadar

Home Depot replaces phone menus with Gemini voice agent that resolves calls 4x faster, expanding nationwide after 50-store pilot - New York Post

Scrunch study identifies four AI search user intents, finds AI referrals more engaged than traditional search - Fast Company

Ex-OpenSea CTO launches Noscroll, a $9.99/month AI bot that doomscrolls feeds and sends personalized summaries and alerts - TechCrunch

Scan your retina before swiping right: Tinder partners with Sam Altman's World on iris-scan proof-of-humanity badges - Dexerto

AI influencers raking it in - New York Post

Follow the latest AI headlines via druce.ai on Bluesky