Newsletter
Five minutes. What happened, and what to do about it.
ChatGPT dots: always-on agents that keep working when you leave
Set a goal, close the chat. The dot carries on in the background.
Office of the Governor of CaliforniaCalifornia: AI can no longer fire or discipline staff on its own
GoogleGoogle announces Gemini 4 Argon, but only cyber defenders get it first
EuronewsAI firms sign a voluntary White House pledge to police themselves
Claims Journal (Reuters)US appeals court: copying to train a rival AI tool is not fair use
OpenClaw Enterprise: Red Hat, Nvidia and OpenAI try to make agents safe for work
The Register· 30 SepWork has started on OpenClaw Enterprise 1.0, a business version of the open-source OpenClaw agent. It adds separate spaces for different teams or customers, hard security boundaries, a control panel for agents and audit trails across an agent's life. Red Hat, Nvidia and OpenAI are behind it. Gartner had called OpenClaw an unacceptable cybersecurity risk. No release date is set.
- If staff run OpenClaw or similar agents on work laptops, find out now and decide whether to allow, restrict or ban it.
- When a vendor says an agent is enterprise ready, ask for the audit log and the permission settings before you believe it.
Source: The Register. This is our short summary of their reporting. The commentary in it and the things to do are SeedFoundry's opinion. Picture from the source, unless it is our own graphic. How we source and credit
Help Net SecuritySecurity tools can now watch what staff share with Claude
Help Net SecurityAn Irish firm launches AI-assisted cyber checks for small businesses
Help Net Security92% of serious flaws in UK firms stay open for more than 90 days
CNBCThe FTC opens an investigation into OpenAI, Anthropic and other AI firms
TechCrunchReddit is ending RSS feeds and free API access because of AI scraping
TechCrunchDoorDash launches an AI agent you order food from by text message
EngadgetMeta's Edits app gets an AI assistant that reads your Instagram numbers
TechCrunchMeta denies that its Muse agent read a user's messages without permission
The Next WebMistral's boss says the US safety debate covers for rivals' negligence
Help Net SecurityGenea lets AI agents manage who can open which door
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