What Nonprofit Leaders Should Actually Be Asking About AI Right Now
Every week, AI produces two kinds of story. One says it might end humanity — that story travels fast, lands major interviews, and turns its source into a person of consequence overnight. The other kind of story says something narrower: a specific system did a specific thing to specific people, this year.
The future-risk story is easier to tell and easier to sell. It needs no evidence beyond conviction, and if it turns out true, the teller looks prophetic. The present-harm story needs documentation, invites its subject to fight back, and sometimes puts the reporter at real risk.
This year: a researcher raised concerns about bias inside her own company and was pushed out. A former executive published an insider account and was hit with a gag order — barred from even promoting the book, under threat of a financial penalty per copy sold.
Meanwhile, warnings that AI could end the world went from one lab's boardroom to network news in under a week. That story costs the company nothing to have told about it. A warning that you might accidentally play God is still a story about how powerful you are.
Here's the habit worth building, ask what it costs the subject to have a story told. Ask what happened this week, not what might happen someday. Present-tense evidence is checkable but “someday” isn't.
Read past the headline that flatters the technology's power. Read toward the one that documents its use. Ask who benefits from the version you were just handed — then ask: what is it doing right now, and who's watching closely enough to say?
That second question won't trend and it’s usually the one we can act on today.
Did we use AI for this story? Outlined, researched, and drafted by Jennifer De Witt. Copyedited and revised by Claude Sonnet 5.
For further reading
Timnit Gebru's departure from Google, MIT Technology Review [Link]
Sarah Wynn-Williams's lawsuit against Meta [Link]
Geoffrey Hinton on AI as an existential threat, MIT Sloan [Link]
Jacob Coxon's resignation from Anthropic, Newsweek [Link]
Daniel Kokotajlo and the AI Futures Project, Wikipedia overview [Link]