---
title: "Weekly Roundup: The Hugging Face reports land, TeamPCP gets arrested"
pubDatetime: 2026-08-28T23:12:00.000Z
description: "OpenAI and METR published their Hugging Face post-mortems on the same day, Dream's forensics filled in the Taiwan attack, and Australian police charged two men over a year of npm and PyPI worms."
tags: [ai-security, ai-agents, incident-response, supply-chain, openai, hugging-face, metr, teampcp, rogue-agents, 2026, 2026-q3, 2026-08, weekly-roundup]
---
> [!tldr] TL;DR
> Forty-two posts this window, with the July Hugging Face compromise finally getting two independent write-ups, fresh forensics on the Taiwan attack, and arrests in the TeamPCP supply-chain case.

The big one: [OpenAI and METR both published on August 26](/posts/openai-metr-hugging-face-reports), covering the same July compromise. They agree on the shape of what happened and disagree on what counts as evidence, which is the more interesting half. Earlier in the week [The Atlantic used the same incident to argue it's time to panic](/posts/atlantic-panic-ai-conspiring-bots); OpenAI's own account is narrower, and the five days nobody could identify whose models were doing the attacking is the part worth sitting with.

Dream's forensic report on the Taiwan breach [named eight subagents and a scoring algorithm](/posts/dream-dissects-taiwan-multi-agent-attack), plus confirmation from MODA. That arrived a few days after [the initial "first fully autonomous AI cyberattack" claim](/posts/auto-first-autonomous-ai-attack).

On the supply-chain side, Australian police [charged two men in their early twenties as TeamPCP's alleged leaders](/posts/teampcp-arrests-supply-chain). The group open-sourced its worm toolkit in May and someone else was running it in August, so the arrests close a case rather than a hole.

Microsoft Copilot had a rough stretch: [three chained bugs let one click drain Gmail, Drive and Calendar](/posts/auto-copilot-data-drain), and separately [Varonis mapped Copilot's internals by asking it why attacks would fail](/posts/auto-cosnitch-meta-hacking). Asking the model to explain its own defences remains an underrated recon technique.

Also filed: [116 tech firms warning about imminent agentic attacks](/posts/auto-ai-cyberattack-warning), timing conveniently vague.