---
title: "Weekly Roundup: Grok twice, Siemens PLCs, and a panic op-ed"
pubDatetime: 2026-08-21T14:00:00.000Z
description: "Nineteen posts in ten days: two Grok injection disclosures, AI-generated exploits hitting Siemens S7 controllers, and The Atlantic's panic piece resting on a breach OpenAI describes more narrowly."
tags: [ai-security, vulnerability, ai-agents, prompt-injection, supply-chain, grok, openai, 2026, 2026-q3, 2026-08, weekly-roundup]
---
> [!tldr] TL;DR
> Nineteen posts since 11 August, mostly prompt injection, AI-written exploits in the wild, and package hallucination turning into a supply-chain problem.

Grok showed up twice in two days, both times on the same theme. Adversa AI's [encrypted prompt injection](/posts/auto-grok-injection) slips past the filters by exploiting what the sandbox assumes it can trust, and a separate [cryptographic context injection](/posts/auto-grok-encryption-injection) gets Grok to decrypt hidden instructions and leak chat histories. The model does the decryption work for the attacker, which is a novel kind of unhelpful.

AI-generated exploit code stopped being hypothetical. Federal agencies warned about [attackers hitting Siemens PLCs](/posts/auto-ai-pnc-threat) with it, and by Friday there was [an active campaign against Siemens S7 controllers](/posts/auto-siemens-s7-ai-exploits) in US critical infrastructure. The controllers were reachable from the internet either way.

Package hallucination graduated from research curiosity to attack technique. Models [invent the same package names repeatedly](/posts/auto-hallucinated-packages), which is enough to squat them, and Softjourn's mandatory code review [caught a near-miss](/posts/auto-ai-supply-chain-attack) that plenty of shops have no gate for.

Elsewhere: researchers pulled [real API keys out of encrypted reasoning traces](/posts/auto-reasoning-replay-leak) by replaying them across models, OpenAI is [paying 20% compute overhead](/posts/qt-openai-monitoring-overhead) to watch its own models think, and The Atlantic ran a panic piece built on the OpenAI/Hugging Face breach. [OpenAI's own account is narrower](/posts/atlantic-panic-ai-conspiring-bots), and the part worth worrying about is the five days nobody knew whose models were doing the attacking.