AI In Law — Episode Summary

Episode Overview

Episode Title: What You Need to Know Before Using DeepSeek AI
Host: Mitch Jackson
Date: January 30, 2025

This episode of "AI In Law" centers on the critical, often-overlooked risks of using DeepSeek AI in legal practice. Drawing from a recent post by legal tech expert Mitch Jackson, the hosts break down alarming issues found in DeepSeek’s Terms of Service (TOS). The discussion covers data retention, surveillance, international legal exposure, and DeepSeek’s unilateral power to change its policies, aiming to arm legal professionals with the knowledge necessary for safe AI adoption.

Key Discussion Points & Insights

1. The Realities Hidden in DeepSeek’s Terms of Service

• Most users agree to TOS without reading them, putting themselves at risk.
• In the legal field, informed consent about data usage is crucial, especially when handling confidential information.
• "Most people don't realize that the TOS, those are like the contract between you and the company." — Speaker B [01:01]

2. Data Retention: Losing Control Over Your Information

• DeepSeek’s TOS states they may retain your data indefinitely—even after account deletion.
• This means any prompts, documents, or information you input could be stored—and potentially monetized—forever.
• "They can keep your data indefinitely." — Speaker B [02:10]
• "So they could be making money off our data even after we stop using their service?" — Speaker A
• "Exactly. It's about control, you know, and as a user, that lack of control should be a big concern." — Speaker B [02:41]

3. Surveillance and User Monitoring

• DeepSeek reserves the right to monitor and collect everything you input and everything the AI outputs, including keystrokes and editing history.
• Usage of the service for sensitive legal, financial, or business brainstorming may leave your confidential data exposed.
• "According to the TOS, they can collect all of that, even down to your keystrokes and editing history." — Speaker B [03:34]
• "It's like they're looking over your shoulder constantly." — Speaker A [03:36]

4. Legal Exposure: Foreign Jurisdiction and User Risks

• DeepSeek is governed by Chinese law, meaning Chinese authorities could potentially demand access to your data.
• U.S. or EU legal protections may not apply, and recourse for international users is limited and unclear.
• "That raises a whole other set of questions, especially for users outside of China." — Speaker B [04:05]

5. The Moving Goalposts: Unilateral Changes to Terms

• DeepSeek can change its TOS at any time, without notifying users.
• This increases legal and ethical risks, as rules and data practices could shift unexpectedly.
• "So the rules of the game could completely change, and we wouldn't even know about it." — Speaker A [04:36]
• "There's not much you can do about it except stop using the service." — Speaker B [04:58]

6. The Bottom Line: Who Really Controls the AI?

• Users should realize that with some AI services, "You're not just using the AI, the AI is using you." — Speaker B quoting Mitch Jackson [05:05]
• The power imbalance between users and AI providers, especially around data and privacy, demands vigilance and skepticism.
• "It feels like we're handing over control of our data and potentially even our privacy without fully understanding the consequences." — Speaker A [04:45]

Notable Quotes & Memorable Moments

• [01:13] Speaker B: "With anything like that, especially when your data is involved, understanding those terms is absolutely crucial... It's about informed consent."
• [02:13] Speaker A & B:
  • "Indefinitely?"
  • "Indefinitely."
• [03:12] Speaker B: "Their TOS... gives them a pretty broad power to monitor and collect everything you put into the app and everything it generates for you."
• [03:36] Speaker A: "It's like they're looking over your shoulder constantly."
• [04:00] Speaker B: "That means Chinese authorities could possibly potentially demand access to user data."
• [05:05] Speaker B: "Mitch really hit the nail on the head in his post. He said, 'You're not just using the AI, the AI is using you.'"
• [05:41] Speaker A: "Your privacy is at stake."
• [05:42] Speaker B: "You have to ask yourself, what level of privacy am I willing to give up? How important is data security to me? Am I comfortable with the potential legal implications?"

Timestamps for Key Segments

• [01:01] — Why TOS matter in legal tech and the basis for "informed consent"
• [02:05] — DeepSeek’s data retention policy: data kept indefinitely
• [03:02] — Surveillance & monitoring: scope of DeepSeek's data collection
• [03:56] — Legal exposure: implications of DeepSeek's TOS being governed by Chinese law
• [04:35] — Powerless users: unilateral TOS changes
• [05:05] — "You're not just using the AI, the AI is using you."
• [05:41] — Action steps for evaluating AI tools and data privacy

Final Takeaways

• Evaluate AI Tools Thoroughly: Don’t let tech hype outweigh privacy and security risks. Read the TOS—especially for tools handling sensitive legal data.
• Data Stays Behind: Even when you leave, your information might remain with the company, out of your control.
• International Law Creates New Risks: Services governed by foreign law may put you and your clients’ data at unique risk.
• Stay Informed: As AI becomes a bigger part of daily legal work, pause before agreeing to terms—your practice and your clients may depend on it.

"Be informed, be aware, and always read the fine print." — Speaker A [06:17]

Recommendation:
Connect with Mitch Jackson on LinkedIn and Bluesky for ongoing updates on AI law and legal tech best practices.