Last week, I was half-listening to one of my favorite podcasts during a workout when something OpenAI CEO Sam Altman said immediately caught my attention. He was chatting it up with comedian Theo Von on This Past Weekend when he made a casual remark that should raise serious concerns for anyone who’s ever shared something personal or legally sensitive with an AI platform (like ChatGPT or Google’s Gemini).
What did he say? In reference to user chats with ChatGPT, Altman remarked: “We could be required to produce that.”
Translation? If your conversations with AI ever become relevant to a legal dispute, there’s a real possibility they could be subject to discovery.
The Illusion of Privacy
This hit me hard because let’s be real here: people don’t just use ChatGPT, Claude, and Gemini for fun facts, or grocery lists. Many rely on it as a digital diary, an informal therapist, an HR consultant, a breakup coach, and even a makeshift legal advisor.
Consider a few real-world examples of prompts:
- “Write a professional email telling my boss I didn’t mean to sound passive-aggressive (but she kinda deserved it)”
- “Help me explain to my ex why what happened in Vegas technically wasn’t a lie.”
- “Re-write this cease-and-desist letter to sound scarier but like without crossing the legal line and make it sound like a real lawyer wrote it”
- “How do I tell my business partner he’s an idiot without saying he’s an idiot?”
These seem harmless, right? But in a dispute, they could become fair game. AI platforms aren’t lawyers, doctors, or fiduciaries. They’re not bound by any duty of confidentiality. What you type may be stored, reviewed, or even disclosed under subpoena. There’s no recognized legal privilege.
An Emerging Regulatory Landscape
Generative AI is new and largely unregulated, but that’s changing fast. On April 18, 2025, Arizona Governor Katie Hobbs signed H.B. 2342, prohibiting local governments from restricting lawful use of “computational power,” explicitly including AI and other resource-intensive computing tasks.
Then, on July 4, 2025, President Trump signed the One Big Beautiful Bill Act, which:
- Bars certain foreign entities from the AI supply chain
- Imposes broad extraterritorial rules on “prohibited foreign entities”
- Mandates enhanced domestic sourcing and supply-chain integrity
On July 23, 2025, the White House released an AI Action Plan advancing its “global AI dominance” policy, signaling even more federal focus on AI governance and compliance.
Together, these laws and plans show AI is rapidly moving from a regulatory gray zone to a tightly controlled space, with increasing legal and compliance risks.
High Stakes for Professionals
Lawyers face added risk. According to the ABA, nearly half of all attorneys expect AI to be mainstream in legal practice within three years. Yet uploading confidential client info into a public AI tool may violate professional rules (like Rules 1.4 and 1.6). Delegating substantive work to AI without review can trigger malpractice or insurance coverage issues especially with growing “Silent AI” gaps in LPL (Lawyers’ Professional Liability) policies.
A Simple Rule of Thumb
If you wouldn’t want something read aloud in a deposition, don’t type it into AI. That includes:
- Legal or client communications
- HR and personnel issues
- Business strategy or trade secrets
- Personal disputes or sensitive matters
Use AI Wisely, But Know Its Limits
AI is excellent for drafting, brainstorming, and research. But it’s not confidential, protected, or bound by any duty of loyalty.
So, if it’s serious (e.g., legal, personal, professional) please talk to a real human. A lawyer. A therapist. A colleague. Someone who’s legally and ethically obligated to protect what you share.
Bottom line: Use AI for what it’s good at. But please remember it’s not your lawyer.
