Let’s Womansplain
▶ LISTEN: “CAN I HANG UP?” (IT'S AT 6:13)
Your personal ChatGPT logs can be discoverable. Talking to a chatbot isn't like talking to your attorney.
That was Logan's warning to Ashley, who wrote in asking whether she can use her own AI at home for work thinking when work says no. It's not hypothetical: in February a federal judge ruled that a fraud defendant's chats with Claude about his case weren't privileged, the first ruling of its kind. The ruling →
Ashley's was the first Let's Womansplain, our new listener segment. The rule we landed on is below, and there's a checklist you can paste into your notes.
INSIDE THE EPISODE
Four things for you to take away
01 · Brainstorm at home. Bring the framework to work. Never the reverse. Ashley’s company won't let her use AI at work. Mara's first move: read the stated policy and know what it says. If there isn't one, or it's unclear, a private account with the privacy controls on is where she'd draw the line. Logan agreed on the policy, then added the warning up top.
Logan flipped the direction of travel: take your own expertise and lived experience into your personal AI, brainstorm there, then carry what you built back into your work context. One thing we'd add. Your employer's policy still applies, and a personal account doesn't override it. If the policy is missing or unclear, ask what's permitted, and put work details only into a tool that's approved for it. Ashley's question, at 23:29 →
THE DOC · FREE, NO SIGNUP
Before you paste: a personal AI checklist
Start with your policy. Then Wanjiku's traffic light, four questions to ask before you paste, a before and after you can copy, and a note to your manager if there's no policy. Take it, share it, paste it into your own notes.
02 · Green is anything you'd put on a billboard. The traffic light is Wanjiku Kamau's. She was our first guest, and on September 12 she ran a session on personal AI use at the AI-Powered Women conference at MIT that Logan sat in on. Red never goes in your own AI: anything from your company, plus whatever you decide is off limits for you, like finances or health. Yellow means pause. Taking names out isn't enough if what's left still gives away something confidential or makes the situation recognizable. Green is anything you'd be fine seeing on a billboard with your name on it.
It works because it makes you stop before you paste. Mara's follow-up was the better question: what's the difference between telling a machine and telling a friend? Logan's answer is that the machine keeps a record, and with wearables and meeting note-takers, so does the friend now. Wanjiku's episode → · Follow her on LinkedIn → · Her book, Out of the Loop, Into the Algorithm → · Her AI-Powered Women speaker card →
03 · Who rated themselves higher? A new Corrix study scored 885 AI working sessions, roughly 37,100 messages. Before they started, people rated their own AI ability. Logan asked Mara to guess which way it went. Mara: "Is this a trick question? … Don't we know the answer to this already?"
Men rated themselves higher than women did. Women scored higher on average in the actual sessions. And years of use, how often people used AI, and company-provided training weren't linked to better work in this study. Logan's read on why: the models keep changing, and once you're comfortable you stop challenging the output. Mara's word for it: complacent. The study →
04 · Data-center protests are a red herring for stopping AI. Mara's appetizer was the week AI safety went mainstream: a researcher resigning from Anthropic over "gambling with our lives," a colleague saying they earnestly believe AI could kill all humans, and one OpenAI researcher's estimate of a 70 percent extinction risk without regulation or a coordinated slowdown. All of it landed on the cover of TIME.
Her way through it was Michael Noetel's piece in The Conversation. His first point is the one she wanted on record: protesting a data center over local water and power is a reasonable fight to have through local democracy, but as a way to slow AI it's misguided. Two-thirds of Americans say AI is moving too fast. The plan Noetel wants ready before a crisis is delays, transparency, and verification, and Mara's Fed years say the verification part means examiners sitting inside the labs. Noetel's piece → · The TIME cover story →
Anybody who's made it this far to the end of our dinner: you are not behind.
LET'S WOMANSPLAIN
Using your own AI when work says no? Your turn.
Ashley's was the first. It aired Thursday, worked through on air, first name only. Yours is next. Anonymous if you want.
This week's question:
What's the AI question you've been too embarrassed to ask out loud at work?
Or hit reply and type it, or email [email protected] any time. A sentence is plenty.
LISTEN
Listen to the episode
Want to hear the whole conversation? Episode 29, Can I Hang Up?, is 36 minutes with Mara Bolis and Logan Currie, recorded while Mara was in Riga.
Short on time? The TIME quotes start at 3:46, "can I hang up" at 6:13, the real-life Olivia Pope at 13:12, Ashley's question at 23:29, the traffic light at 27:19, the study at 29:55, and "you are not behind" at 34:10.
One more thing from the conversation:
The A.I. Mob That Attacked Hugging Face
The Hard Fork episode with METR's Ajeya Cotra that gave Logan physical chills. Her worry number one comes straight from it. Listen →
NEXT EPISODE · THURSDAY, OCTOBER 1
Bella Pivo
23, a pre-seed VC analyst, 100,000 followers on TikTok, and not sold either. Plus a digital garden you can start in an afternoon.
WRITTEN BY
Logan Currie & Mara Bolis
Women, work, and the messy middle of AI.
P.S. Mara's estrogen made it home. The suitcase turned up the day we recorded, which is the real ending of the episode.
