Here's the uncomfortable part of the AI-at-work boom: the tool you trust is often a thin layer on top of other companies. This week, Notion told a small set of customers that their data was compromised in a breach at Fireworks AI, a cloud platform that runs open-weight models for apps like Notion.
What happened
According to SecurityWeek, Fireworks was hacked twice in October. The second incident, on October 24, involved an unauthorized workload and was linked to a rise in LLMJacking. Fireworks contained it the same day and told Notion on November 2 that Notion AI data had been accessed.
The exposed material was chat transcripts from the Insights feature, which helps customers see how they use AI. Fewer than 1% of Notion workspaces were affected, and Notion says most transcripts can't be tied back to a specific user, workspace or company.
Why it matters for your career
Think about what you've typed into a workplace AI box: draft performance reviews, client names, a stray API key, a salary spreadsheet. A transcript doesn't need a name attached to be sensitive if the content itself gives it away.
Notion's advice: treat the data as exposed, and rotate any passwords, API keys or credentials pasted into Notion AI.
- Rotate any secret you've ever pasted into an AI tool.
- Ask your employer which model providers sit behind your AI tools.
- Keep client-identifying details out of prompts when you can.
Timing is everything, and the time to build AI hygiene is before the next notification lands in your inbox. Source: SecurityWeek.
