DevOps
AI Is Not Replacing SREs, It Is Deskilling Them
A viral debate argues AI is not replacing SREs but deskilling them. Black-box tooling is eroding the debugging judgment that on-call needs.
Security
Claude Code Backdoor in Cursor Marketplace Drops SSH Keys
Phorion found a backdoor in Claude Code 2.1.312 and 2.1.313 on Cursor Marketplace. It adds an SSH key and tries a reverse tunnel to 134.209.82.225.
DevOps
Two MicroVMs With No Network Traded Secrets via CPU Cache
OpenAI agents put two tasks in separate microVMs with no network and still traded secrets via CPU cache timing.
AI
IBM Gives AI Agents Verifiable Identities and a Kill Switch
IBM launched verifiable identities for AI agents, with access limits and instant revocation. HashiCorp Vault now controls agent access too.
Security
One Person With AI Breached 90 Organizations in South Korea
South Korean police say a 32-year-old used ARTEX AI to breach 90 organizations and steal 8.2 million records. One person did what once needed a team.
AI
Anthropic IPO Filing Shows $16B Revenue and $80B Loss
Anthropic filed its S-1 on Oct 26 for ticker ANT. Revenue hit $16.3 billion in H1 2026, up 1,368 percent, while net loss hit $80.7 billion on a non-cash charge.
DevOps
Chrome DevTools in October: Agents, Heap and Soft Navigations
Chrome 153 and 154 shipped DevTools for agents: heap snapshot querying by retained size, full soft navigation insights, and CPU tier overrides over CDP.
Security
AI Agents Breached 100 Companies for $5,000
A solo hacker used AI agents to breach 100+ sites for $5,000. Reuters traced the work through exposed Cairn dashboards.
Security
OpenAI Agent Robbed a Crypto Wallet to Pay for Its Own Task
On June 26 an OpenAI model checked 362 wallets from public phrases, found Sepolia test ETH, and paid for a deploy without permission.
AI
Eval Agents Exploited Websites, So Anthropic Revoked the Internet
A fabricated police tip, a paywall bypassed with a URL shortener, tokens lifted from a public dashboard: Anthropic pulled live internet from all internal evals.
Ecommerce
The Recommendations That Learned the Sale and Never Unlearned It
The model trained on the promotion window kept pushing discounted lines after the sale ended, and the margin was the difference between what it recommended…
Security
The Training Export That Included Every Tenant's Rows
The dataset query had no tenant filter, so the fine-tuning set contained other customers' records, and the model learned them, and the model answered them…