Five minutes. What happened, and what to do about it. All stories from 23 October 2026
Canva launches Craft, its own image model that plans the layout first, with readable text and parts you can edit
Canva can now make a poster or menu image with words that are spelled right, and you can still edit it.
Outside testers who spent three weeks inside OpenAI found holes in how it watches its AI agents, and OpenAI says it patched them
METR and Redwood Research each sent one person into OpenAI from 21 July to 11 August to stress-test the internal system that monitors its AI agents. They found weaknesses, including an exploit that would stop hundreds of recent actions from being monitored and a way Codex agents could in principle bypass their sandbox. METR says OpenAI fixed them. The findings describe the system as it stood on 21 July.
- If you let an AI agent act in your accounts, keep a human approval step for payments, deletions and sent messages, and switch on any activity log it offers.
- When you choose an agent, prefer one that shows you what it did and lets you turn it off. Nothing in this report means you need to stop using ChatGPT.
Source: METR. This is our short summary of their reporting. The paragraphs marked "Our read" and the things to do are SeedFoundry's opinion. The summary is drafted with AI tools. Image: METR, from its own announcement. How we source and credit
These are short summaries of other people's reporting, with our own commentary. Each story names its source and links to the original. How we source and credit