AI news

Five minutes. What happened, and what to do about it. All stories from 7 November 2026

HIGHLIGHT

AI data brokers are buying old company emails, chats and files to train models, and a sale of Spirit Airlines records is on hold over staff privacy

The emails and chats you sent at work may one day be sold to train an AI, and the firms buying them say the names are stripped out.

WIRED|6 NOV

A Vals test finds free, downloadable AI models nearly match the best closed ones at hacking mock networks, at a fraction of the cost

WIRED· 6 Nov

WIRED reports that Vals, a company that measures AI abilities, tested models on mock corporate networks. OpenAI's cyber model Daybreak, available only to trusted partners, completed 81% of tasks and the open-weight model DeepSeek completed 73%. Vals says an attack cost $17 to $46 in tokens with GPT-6-Sol and $2 to $3 with DeepSeek. Open models can be downloaded and run on one's own hardware.

  • Turn on two-step sign-in or passkeys for email, banking and your website host, and install updates as soon as they appear, since this raises the odds that weak setups get found.
  • Ask whoever looks after your IT how quickly they patch, and which of your systems face the internet.

Source: WIRED. 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: Vals AI, from its own announcement. How we source and credit

That is sat 7 nov, all of it.

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