Proactive WAF Vulnerability Protection & Firewall for AI + Multiplayer Chess Demo in ChatGPT

#ThisWeekinNET — Episode 112

In this episode of This Week in NET, we talk with Daniele Molteni, Director of Product Management for Cloudflare’s WAF, about how Cloudflare responded within hours to a newly disclosed React Server Components vulnerability — deploying global protection before the public advisory was even released.

That speed matters. In just the first 11 days after disclosure, Cloudflare observed more than 1 billion exploitation attempts related to React2Shell, with sustained pressure averaging over 4 million hits per hour, and peaks far higher. Threat actors quickly integrated the vulnerability into large-scale scanning and reconnaissance, targeting even critical infrastructure. If you run React, upgrading is urgent.

Daniele explains how WAF rules are built, how new payload logging improvements help customers understand real attack traffic, and what’s coming next in 2026 — including Firewall for AI, fraud detection, and safer, gradual rule rollouts.

To close the episode, Systems Engineer Steve James gives a hands-on demo of a real-time multiplayer chess app running inside ChatGPT, built with the Agents SDK and Cloudflare Workers.

⏱️ Timestamps

02:13 — The Cloudflare WAF and the new React Server Components vulnerability

07:15 — How Cloudflare deployed protection early (Emergency Release)

08:30 — Payload logging improvements and encrypted visibility

11:00 — Looking ahead to 2026: Firewall for AI, fraud detection, rule rollout

12:17 — What makes LLM-specific attacks different (jailbreaks, prompt injections)

14:10 — Demo: Steve James on building apps with the Agents SDK; Multiplayer real-time chess game inside ChatGPT

18:30 — Asking the AI for chess analysis mid-game

21:10 — How the app works (MCP servers, Workers, real-time sync)

22:00 — The future of agent ecosystems & why Workers is ideal

Subscribe for more weekly conversations on Internet trends, infrastructure, and technology:
ThisWeekinNET.com