Subscribe
, ,

What the Notion AI chat leak means for you, even if you never use Notion

AIooze Editorial

In one minute

  • What changed: a company that runs AI models for other businesses was broken into, and some Notion AI chat transcripts were exposed.
  • Who it affects: fewer than 1% of Notion workspaces, according to Notion, which says it has contacted them. The lesson applies to every chatbot user.
  • What to do: change any password or key you have ever pasted into an AI chat, then spend 10 minutes on your chatbot settings.

What happened

Fireworks AI runs AI models on behalf of other companies. On 9 November 2026 it said attackers had reached more customer data than it first reported. Notion used Fireworks for some of its AI features. According to Notion’s security advisory, the exposed data “included Notion AI chat transcripts used for the Insights feature”, affecting fewer than 1% of workspaces. Notion says it cut the connection on 3 November and began contacting affected customers on 9 November. If you have not heard from Notion, it says your workspace was not identified as affected. Our news brief has the full story.

Why it matters beyond Notion

When you type into an AI tool, your words can pass through other companies that run the models behind it. You will rarely see their names. A break-in at any of them can expose what you typed, even if the app you use was never hacked itself.

What to do

  1. If Notion has emailed you, follow its instructions to review the affected data inside your workspace.
  2. Notion’s advice to affected customers: if you pasted passwords, API keys or other sensitive details into Notion AI during the incident window, change them. We suggest going further: if you have ever pasted a password or key into any chatbot, change it now.
  3. Make it a rule: no passwords, bank details, ID numbers or other people’s private information in any AI chat.
  4. Decide whether your chats may be used to improve the company’s models. As of 12 Nov 2026: OpenAI’s Data Controls FAQ says ChatGPT’s setting is “Improve the model for everyone” under Settings, then Data Controls. Google’s Gemini privacy hub says turning off Keep Activity stops future chats being reviewed to improve its services. Anthropic’s privacy centre says Claude’s setting is “Help improve our AI models” under Settings, then Privacy.
  5. Delete old chats you no longer need, especially any with work or personal details in them.
  6. Expect phishing. After any leak, criminals may send convincing “security” emails. Do not click their links; open the app or website yourself.

Watch out for

Turning off training does not mean nothing is kept. Google, for example, says chats with Keep Activity off are still kept for 72 hours. Check each maker’s help page for how long copies are held.

Go deeper

Sources

Related