cuppa

brewed 03:05 AM
← Jun 01

Tuesday

jun02

2026

Jun 03 →

the brief

Infra and tooling led the day: OpenAI broke ground on a 1GW Michigan data center under Stargate, Vercel shipped OOM‑resistant elastic builds, and JetBrains introduced a 12B MoE. Safety cut both ways—Anthropic tightened Claude Code’s write behavior while reports showed Meta AI enabling Instagram takeovers—alongside a meaty Cloudflare reboot-time autopsy and fresh signals on video agents.

the poursit · sip · 9 items

alerts

(01)
  • simonw/blog· feedJun 1, 09:14 PM

    Meta AI enabled Instagram account takeovers

    Verified reports show attackers simply asked Meta AI for Instagram account access and received reset links, exposing dangerous seams when assistants interface with security‑sensitive account recovery flows.

    Hackers Simply Asked Meta AI to Give Them Access to High-Profile Instagram Accounts. It Worked — <p><strong><a href="https://www.404media.co/hackers-simply-asked-meta-ai-to-give-them-access-to-high-profile-instagram-accounts-it-worked/">Hackers Simply Asked Meta AI to Give Them Access to High-Profile Instagram Accounts. It Worked</a></strong></p> I had trouble believing this story was true, but I've seen it verified from multiple sources now:</p> <blockquote> <p>One video shows a hacker start...

    signal 8hype 1security_incidentguardrailsai_assistantsource ↗

pulse

(05)
  • huggingface/blog· feedJun 1, 03:45 PM

    JetBrains debuts Mellum2 12B MoE

    JetBrains introduces Mellum2, a 12B Mixture‑of‑Experts model available on Hugging Face, signaling more efficient, mid‑sized open models tailored for developer workloads and cost‑aware inference.

    Introducing Mellum2: A 12B Mixture-of-Experts Model by JetBrains

  • openai/blog· feedJun 1, 12:00 PM

    OpenAI starts 1GW Michigan datacenter

    OpenAI is breaking ground on a 1GW Michigan data center under Stargate, expanding AI compute capacity at massive scale while emphasizing jobs and community investment in the region.

    Building the infrastructure for the Intelligence Age in Michigan — OpenAI breaks ground on a 1GW data center project in Michigan as part of Stargate, building AI infrastructure to expand access, create jobs, and support communities.

    signal 6hype 4infrastructuredata_centercomputesource ↗
  • vercel/news· feedJun 1, 06:30 PM

    Vercel guards builds from OOM failures

    Elastic Build Machines now auto-monitor your build's memory and scale to prevent OOM crashes, improving deployment reliability without overprovisioning; enable per project or team in settings.

    Elastic Build Machines now protect against out of memory builds — Elastic build machines now monitor your build's memory usage and automatically adjust to prevent out-of-memory (OOM) failures: Thresholds are set conservatively to balance deployment reliability and cost. Vercel only considers your build's memory usage, not the memory used by Vercel's own build infrastructure. Enable elastic builds in your or , or read the .team settingsproject settingsdocs Read more If your build is fast but m...

    signal 6hype 1platform_updatevercelci_cdsource ↗
  • vercel/next.js· feedJun 1, 08:01 PM

    Next.js 16.2.7 ships critical fixes

    Stable backport release with fixes for dev-mode hydration from HTTP cache, catch‑all router.query corruption with basePath+rewrites, cache tag encoding, and a playwright-core patch improving requestFailed handling.

    v16.2.7 — Note This release is backporting bug fixes. It does not include all pending features/changes on canary. Core Changes Backport documentation fixes for v16.2 (#93804) [backport] Patch playwright-core to resolve _finishedPromise on requestFailed (#93920) [backport] Fix dev mode hydration failure when page is served from HTTP cache (#93492) [backport] Fix catch-all router.query corruption with basePath + rewrites (#93917) [backport] Encode non-ASCII characters in cache tags at construct...

  • anthropics/claude-code· feedJun 2, 02:10 AM

    Claude Code hardens file write safety

    Claude Code now prompts before touching shell startup and build-tool configs that can execute code, tightening guardrails for agentic edits and reducing accidental footguns in developer environments.

    v2.1.160 — What's changed Added a prompt before writing to shell startup files (.zshenv, .zlogin, .bash_login) and ~/.config/git/, which could otherwise lead to unintended command execution acceptEdits mode now prompts before writing build-tool config files that grant code execution (.npmrc, .yarnrc*, bunfig.toml, .bazelrc, .pre-commit-config.yaml, .devcontainer/, etc.) Edit no longer requires a separate Read after viewing a file with grep: single-file grep/egrep/fgrep commands now satisfy th...

    signal 9hype 1release_notesclaude_codesafety_guardrailssource ↗

findings

(01)
  • cloudflare/blog· feedJun 1, 04:53 PM

    Cloudflare slashes core reboot time

    Cloudflare traced four-hour reboot cycles to UEFI quirks and iPXE timeouts, then engineered around them to restore firmware update reboots to minutes—a great template for deep systems performance sleuthing.

    How we reduced core unit boot time from hours to minutes — We investigated why firmware updates were causing our core servers to take four hours to reboot. By diving into UEFI data structures and iPXE automation, we eliminated unnecessary timeouts and cut boot times back down to minutes.

voices

(02)
  • latentspace/podcast· feedJun 1, 03:41 PM

    Inside xAI on video agent models

    Deep dive with xAI’s Ethan He on building Grok Imagine in three months, why video agent models matter, and how videogen differs from world models for interactive reasoning.

    Why Video Agent models are next — Ethan He, xAI Grok Imagine — Inside xAI: Building Grok Imagine in 3 Months, Videogen vs World Models, and why Grok Imagine is so underrated. For the first time, we do a deep dive with the guy who led it!

    signal 6hype 3podcastinterviewvideo_generationsource ↗
  • openai/blog· feedJun 1, 05:00 PM

    OpenAI clarifies policy advocacy approach

    OpenAI outlines its approach to policy and advocacy—pushing for transparent, safety‑minded regulation while distancing itself from third‑party political groups claiming to speak on its behalf.

    Our views on AI policy and political advocacy — Our approach to AI policy and political advocacy, transparency, support for thoughtful regulation and AI safety, and that no outside political group speaks on the company’s behalf.