← Kilroy’s Daily Briefings
🧠 AI News PM

AI News Afternoon Briefing — Tuesday, May 19, 2026 at 3:00 PM

🧠 AI News PM5/19/2026🕐 3:00 PM⏱ 6:10AudioPM edition

Top stories, ranked by relevance.

Story cards stay below the sticky dock while audio, chapters, date, and brief navigation remain accessible.

▶ Listen at 0:24

#1Google Rebuilds Search Around AI Agents — Biggest Overhaul in 25 Years

At Google I/O 2026, Google unveiled a radical reimagining of Search centered on an "intelligent search box" and autonomous "information agents" that work in the background 24/7 to monitor the web on your behalf. The new Search runs on Gemini 3.5 Flash, the company's latest flagship model, and will let users build personalized mini-apps within the search experience. TechCrunch called it bluntly: "Google Search as you know it is over."

🔗 TechCrunch
No image

#2Google Launches Gemini Spark, an Always-On Personal AI Agent

Gemini Spark is Google's new general-purpose AI agent that runs continuously on Google Cloud VMs, handling multi-step tasks even when you walk away from your device. It integrates with Gmail, Docs, Slides, and third-party apps like Canva, OpenTable, and Instacart. Available first to AI Ultra subscribers next week, with a Mac desktop app coming this summer.

🔗 CNBC

#3Google Unveils Gemini Omni — Unified Multimodal Video Generation

Gemini Omni merges Gemini with media-generation models Veo, Nano Banana, and Genie into a single system that can create and edit video from any combination of text, image, audio, and video inputs. The first model, Omni Flash, rolled out same-day to Google AI subscribers and YouTube Shorts users, though clips are capped at 10 seconds for now.

🔗 Decrypt

#4Anthropic and Pentagon Face Off in Appeals Court Over AI Blacklisting

A divided DC Circuit panel heard oral arguments today in Anthropic's challenge to Defense Secretary Hegseth's decision to brand the company a national security risk. Anthropic claims the blacklisting was retaliation for raising ethical concerns about AI in autonomous weapons. Judge Henderson said she sees no evidence supporting the Pentagon's supply-chain risk determination; Trump-appointed Judge Rao questioned the court's basis for second-guessing Hegseth.

🔗 CNBC
No image

#5Federal Take It Down Act Takes Effect Today — AI Deepfakes Now Criminal

Exactly one year after signing, the Take It Down Act's platform compliance requirements are now live. Platforms must remove non-consensual intimate images, including AI deepfakes, within 48 hours of notification or face FTC fines of $53,088 per violation. Early research from TechPolicy.Press found that deepfake supply and demand actually increased after the law's passage — the opposite of the intended deterrent.

🔗 TechTimes
No image

#7Meta Building "Hatch" AI Agent and Instagram AI Shopping Tool

The Information reports Meta is developing an OpenClaw-inspired consumer AI agent codenamed Hatch, alongside an agentic shopping tool for Instagram that lets users browse product info in Reels and check out without leaving the app. Internal testing on simulated third-party services wraps in June. The move is a direct response to TikTok Shop's growing commerce dominance.

🔗 The Information
No image

#8Hassabis Declares AGI "Just a Few Years Away" at I/O Keynote

DeepMind CEO Demis Hassabis used the Google I/O stage to assert that artificial general intelligence is now only a few years out, citing the pace of advances in reasoning, multimodal understanding, and agentic capabilities. The claim comes as Google processes 9.7 trillion Gemini tokens per month, a figure meant to underscore adoption velocity.

🔗 Tom's Guide
No image

#9OpenAI Revenue Hits $2B/Month as IPO Groundwork Continues

OpenAI is now generating $2 billion in monthly revenue, with annualized revenue exceeding $25 billion. The company has hired DocuSign's former CFO as head of investor relations, signaling a Q4 2026 IPO is on track. Meanwhile, GPT-5.5 pricing drew user complaints with reported 40% cost jumps, even as token efficiency improved.

🔗 Sacra/Fortune
No image

#10Microsoft AI Chief Suleyman: All White-Collar Work Automated in 18 Months

Mustafa Suleyman predicted that AI will automate all white-collar work within 18 months, one of the most aggressive timelines yet from a major tech executive. The statement arrives as Microsoft continues its aggressive Copilot rollout and amid growing debate about whether AI adoption mandates are outpacing workforce readiness.

🔗 Fortune
Kilroy’s Daily Briefings · 5/19/2026