SHELFCRITTER AI Digest

TOOL

ChatGPT

tooltopic-noteopenai

Overview

ChatGPT is OpenAI‘s consumer chat product and the surface for most of OpenAI’s user-facing model deployments. The product line has forked into tier-specific surfaces (Plus, Pro, Business, Enterprise, Edu) and feature-specific products (ChatGPT Images, ChatGPT Health, ChatGPT 5.5 Pro). In late 2026 ChatGPT’s identity-layer and pricing-model moves — HTML-preview takeover chains, locked chats, and the chat-and-work surface merger — read as the primary pressure points for the product.

Timeline

  • 2026-11-02-AI-Digest — Parameter Labs discloses a full ChatGPT HTML-preview takeover chain: an untrusted iframe → postMessage into MathLive’s virtual keyboard → innerHTML injection in the parent inserts a fresh unsandboxed iframe → that iframe uses the CSP-allowed YouTube oEmbed endpoint to reflect and execute attacker JS → full code exec on chatgpt.com with data exfiltration via an image request. A shared ChatGPT link visited by another signed-in user was sufficient. OpenAI patched and paid $5,000; Simon Willison called the bounty “a bit cheap for a vulnerability this serious.” Same digest: OpenAI is merging the Chat and Work surfaces inside ChatGPT in December, with Plus/Pro/Business tiers sharing one token budget across ChatGPT and Codex — the all-you-can-prompt tier is being retired by design. Flat-rate pricing on the paid tiers retires; the Instant model remains outside the budget on paper.

Key Developments

  1. Parameter Labs HTML-Preview Takeover Chain + $5,000 Bounty (November 2, 2026): Full exploit chain against the artifacts/HTML-preview surface — MathLive postMessage + innerHTML sink + YouTube oEmbed CSP escape — that executes attacker JS on chatgpt.com in a signed-in user’s session, with a shared link alone sufficient as the entry point. Load-bearing takeaway: a chained CSP + DOM-sink + third-party-embed failure, not a prompt-injection story — teams shipping LLM UIs that render model-authored HTML should audit every oEmbed and share-link preview surface, independent of the model’s safety posture.

  2. Chat and Work Surface Merger — Shared Token Budgets Retire Flat-Rate Pricing (November 2, 2026): OpenAI is merging Chat and Work inside ChatGPT in December with Plus, Pro, and Business tiers sharing one token budget across ChatGPT and Codex; heavier tasks (deeper reasoning, longer agent runs) burn that budget faster, while the Instant model remains outside the budget on paper. The pricing surface now explicitly reflects inference cost, converging consumer plans toward the per-task budgeting developers have had to model manually for API-tier work.

See also: OpenAI, Codex, Simon Willison, MOC - Agent Security, MOC - Major Companies, MOC - Developer Tools.