╔═════════════════════════════════════════════════════════════╗
║ ║
║ ▀█▀ ▄▀█ █▄▀ █▀▀ █ ▀█▀ █▀▀ █▀█ █▀█ █▀▄▀█ ║
║ █ █▀█ █ █ ██▄ █ █ █▀ █▀▄ █▄█ █ ▀ █ ║
║ ║
║ ▀█▀ █ █ █▀▀ ▀█▀ █▀█ █▀█ ║
║ █ █▀█ ██▄ █ █▄█ █▀▀ ║
║ ║
╚═════════════════════════════════════════════════════════════╝
// Tech news at terminal velocity
cat 2026-05-18.md
$ cat TLDR.md
▸ • The Vatican teams up with Anthropic to drop an encyclical on AI ethics and human dignity.
▸ • Eric Schmidt gets booed by college grads who aren't thrilled about their AI-dominated job prospects.
▸ • Meta's Llama 4 deep dive reveals a 405B parameter MoE beast that's closing the gap with proprietary models.
Dust just bagged $40M from Sequoia, Snowflake, and Datadog to turn AI agents from solo assistants into team players. Because the only thing better than an AI hallucinating is an AI hallucinating in a shared enterprise workspace.
Pope Leo XIV and Anthropic's Christopher Olah are teaming up to release a document on AI and human dignity. It's a holy alliance that might just put them on a collision course with the US military's AI ambitions.
In a bid to make privacy its killer AI feature, Apple's upcoming iOS 27 Siri revamp will reportedly include auto-deleting chats. Finally, an AI that forgets your embarrassing late-night questions as quickly as you do.
South Korean startup LetinAR is building thumbnail-sized lenses that could become the optical backbone of the upcoming AI glasses era. Because strapping a smartphone to your face needs to look a little more subtle.
Meta's Llama 4 is here, and the 405B parameter flagship is using a Mixture-of-Experts architecture to go toe-to-toe with GPT-5. This breakdown explores how open-source is finally closing the gap with proprietary frontier models, making local deployment of god-tier AI a reality.
Former Google CEO Eric Schmidt tried to hype up AI during a commencement speech at the University of Arizona and was promptly drowned out by boos. Turns out, graduates entering a brutal job market aren't super receptive to hearing how algorithms will do their jobs better.
Raindrop AI just open-sourced Workshop, a local debugging tool for AI agents that integrates directly with Codex. It gives you detailed Claude code traces and tool-call usage so you can actually figure out why your agent went rogue.
Truthmark's new release brings Git-native, branch-scoped workflows for AI coding agents. It adds verifier and writer agents to help keep your repository's truth aligned when AI starts making functional code changes.
Red Hat is rolling out new tools for building and testing AI agents, including isolated sandboxes in Red Hat Desktop. Now you can safely test autonomous agents locally without them accidentally nuking your host OS.
Microsoft is officially retiring Teams' Together Mode, the pandemic-era feature that made it look like you and your coworkers were sitting in a virtual auditorium. We'll miss the illusion of wearing pants in a shared space.
"Zero Parades: For Dead Spies" is out, asking players to question the price of forgiveness in a failed espionage operation. It's scratching that deep, existential itch left behind by Disco Elysium.