Strong wake-up call. I’d add this: employees don’t leak data maliciously—they do it out of convenience, loneliness, or to gain signal faster.
What lands with me is how casually people paste internal specs into ChatGPT to make deadlines. That convenience becomes a compliance disaster overnight.
We need layered responses: tech controls and cultural shifts—teaching teams to question when to ask a bot, not just how. Controls are critical, but so are guardrails and shared norms about what belongs in AI’s sandbox and what doesn’t.
