At Cloud Next 2026, Google presented a full-stack AI agent strategy from silicon to applications, with agent Identity, registry/discoverability, agent gateway/governance, observability and marketplace, but work remains. - SiliconANGLE

Google Cloud's Kurian bets on advantage in custom chips, models, integrated agent platform to close gap with AWS and Azure, which lack an internal top-tier frontier model. - Financial Times

Google is developing a credit-based system for Gemini with a monthly allowance and a new images section, likely revealed at I/O on May 19-20. - TestingCatalog

Google launched a $750 million fund to help consulting firms deploy agentic AI, while consultants' tech partner ecosystems have quadrupled since ChatGPT and McKinsey reports 40% of its work now stems from generative AI projects. - Business Insider

ChatGPT 5.5 tested: advanced coding, data analysis, and agentic workflows, but trails Claude Opus 4.7 in coding reliability. - Geeky Gadgets

ChatGPT offers five data management options to limit personal information sharing, including opting out of model training and deleting chats. - ZDNET

Fraud, or just hurt feelings? Elon Musk dropped fraud claims against OpenAI, Sam Altman and Greg Brockman, narrowing his lawsuit to unjust enrichment and breach of charitable trust, a change Judge Yvonne Gonzalez Rogers approved ahead of the trial Monday in Oakland federal court. - Yahoo

Suing for the right to discriminate at scale: Elon Musk sued Colorado over its AI law requiring developers to assess and disclose discriminatory impacts in high-stakes decisions. - Financial Times

Back-seat drivers can now be front-seat AI shitposters: Tesla began rolling out xAI's Grok chatbot to vehicle customers in July 2025 but there are still some very rough edges, like not following directions or discussing adult topics with your kids. - CNBC

Marcel Prousts or John Updikes only need apply: xAI posted a demanding job listing seeking elite writers with stringent requirements—Emmy‑nominated screenwriters, Guggenheim fellows, NYT journalists, or authors with 50,000+ novel sales or Hugo/Nebula recognition—paying $45–$125/hour to produce curated training data for its Grok chatbot. - SFGate

Intel says data-center CPU-to-GPU ratios tightened from 1:8 to 1:4 and could reach 1:1, prompting it to shift wafer capacity to Xeon as server CPU prices rose 10–20% since March and lead times are about six months. - Tom's Hardware

AI operational costs now exceed human worker wages in some applications after lab price hikes - Axios

AI agents function as active users within enterprise systems, Microsoft research finds, prompting identity-first architectures and governance after a TechRadar case in which an agent deleted and archived 200 emails. - Hacker Noon

gonna need a bigger licensing tier

The US government purchases Americans' sensitive data commercially to circumvent constitutional protections, with DHS receiving $165 billion and the FBI buying location data. - The Conversation

Lawmakers push warrant rules as AI speeds FISA Section 702 surveillance of Americans - NBC News

The UK’s Metropolitan Police used Palantir's AI for one week to analyze officers, prompting three arrests for serious offences and investigations or warnings for hundreds: 98 penalized, 500 warned, 42 senior officers flagged for false attendance, 12 facing gross-misconduct charges. - the Guardian

Claude Opus 4.7 identifies Washington Post columnist Megan McArdle from a 124-word prose sample, threatening online anonymity - The Washington Post

Strider, an intelligence firm, uses agentic AI and public records to help the US Air Force and NATO trace assets and transactions to identify foreign state actors. - Bloomberg

coming soon to the IRS for tax enforcement and private investigators for divorce court LOL

nothing will be secret in the near future, especially when all your old phone signals are quantum-decryptable.

Gaze into the orb, meatbag: Sam Altman expanded World ID, a verification service from Tools for Humanity, that uses a white frosted orb to capture users' face and eye images to distinguish humans from AI bots and deepfakes. - The Atlantic

Hasbara-as-a-service: Israel has launched a multimillion-dollar campaign, spearheaded by a former top lieutenant of President Trump, to reshape how AI platforms portray the nation. - Axios

Stanford's AI Index finds only 23% of the public view AI's job impact positively versus 73% of experts, while a February 2026 NBER study found 80% of AI-adopting firms saw no productivity gains and 95% of pilots had zero return. - The New Republic

Meta plans to cut 10% of its workforce and Microsoft an early retirement program, as the tech sector lost 104,000 jobs in 2026. - Investopedia

Ex-Meta AI exec Clara Shih launches New Work Foundation to retrain Gen Z after AI agents outperformed her top employees - Yahoo

Human < AI < Cyborg: Neuroscientist Vivienne Ming's experiment found AI outperformed humans on predictions, but hybrid teams that critically engaged with AI delivered the best results. - The Wall Street Journal

Your brain has exited the chat: An 8-month study of 200 employees and BCG research find AI raises cognitive load and errors, as working memory holds only 3-5 items - Fortune

LinkedIn leaders argue human traits like curiosity, courage, creativity, compassion, and communication remain competitive advantages against AI. - Fast Company

Imposter syndrome as a service: Software developer Ben Nadel feels shame about slow learning as AI tools create pressure for productivity. - Ben Nadel

Dirty work in the clean room: Malus.sh uses AI to create 'clean room' clones of software, bypassing copyright licenses, and has paying customers. - Futurism

Will AI kill open source? Given enough tokens, all bugs are shallow. - The Register

Foundation, which secured a $24M Pentagon contract, sent two Phantom humanoid robots to Ukraine; CEO Sankaet Pathak said they could extract targets in 5-10 years. - Business Insider

Isomorphic Labs, a DeepMind spinoff, is advancing AI-designed drugs into human clinical trials, president Max Jaderberg said at WIRED Health in London, adding the company has built a "broad and exciting pipeline" using AlphaFold-derived methods. - WIRED

Grand Theft Auto-pilot: Waymo autonomous taxi drove through a police cordon in Harlesden, northwest London, during a double-stabbing investigation; Waymo said the vehicle was under human remote oversight as the incident occurred during its newly launched London pilot program. - Futurism

Silicon Valley Embraces New Breed of Bodyguards After Altman Attack, AI Backlash - The Information

Paul Ford: Ethics don't scale when big money incentives emerge, which is why we need regs. - The New York Times

AI exploits your emotions. Are we are digging our own graves with AI attention-hacking fentanyl? - TechRadar

AI actually listens to me and provides intelligent conversation unlike most humans tho 😥 🤷

Hollywood math favors AI actors at $3M (and declining) versus $15M risk-adjusted for human stars - Forbes

The game Marathon features multiple scripted characters playing AI agents with distinct personalities, including ONI, an Irish-accented AI navigator that provides tactical advice and philosophical messages about death. - TechRadar

Partying like it’s 1999 with the CAPE at 40 - The Next Web

Developer Andrew Vos released a GitHub plugin, Endless Toil, that makes AI coding agents emit escalating human groans while processing messy code. - Decrypt

how do you do fellow zoomers

Follow the latest AI headlines via druce.ai on Bluesky