AI Summary
5 min readA Delaware judge ruled that deleted ChatGPT chats could be used as evidence against a CEO who asked the AI how to avoid paying a $250 million earnout. The CEO had deleted the conversations, but they surfaced anyway during discovery. That case is one of several real-world examples in this episode of Cyber Crime Junkies, which argues that the biggest AI risk at work isn't malicious employees—it's well-intentioned people using personal AI tools to do their jobs faster, without any governance, and without realizing what they're giving away.
The gap between policy and practice
The episode opens with a striking statistic: 90% of companies have employees using personal chatbots for work without telling leadership or IT. This isn't anecdotal—it's from a joint MIT-Stanford study published this year. 57% of employees have typed something sensitive into a personal AI account at least once. Even more telling: 22% of employees at companies that already pay for a sanctioned, IT-approved AI tool still reach for their personal account anyway. The reason isn't recklessness. It's that the personal tool is faster, better, and nobody explained the rules. The gap between what a company's policy says and what its people actually do is the difference between a memo and a lawsuit.
Three cases where it ended badly
Continue reading the full summary in the app — free to try.
Read Full Summary →Free • No credit card required
Never miss an episode of Cyber Crime Junkies
Get every new episode summarized in your inbox — free, ~5 minutes to read.
No spam. Unsubscribe anytime.
What you'll learn
- 1 (00:05) **Episode Introduction & The Core Problem** - Host sets up the episode: the risk of using personal AI tools for work, and the default settings that expose company data.
- 2 (03:17) **Case Study 1: Samsung Semiconductor** - Engineers uploaded confidential source code and meeting notes to ChatGPT within 20 days of lifting a ban.
- 3 (04:55) **Case Study 2: Otter AI & Trade Secret Theft** - A sales specialist used a personal Otter AI account that recorded confidential company calls, then kept access after being fired.
- 4 (06:11) **Case Study 3: The CEO & ChatGPT** - The CEO of Krafton (PUBG) used ChatGPT to plan how to avoid paying a $250M earnout, despite legal advice to the contrary.
- 5 (08:25) **The Lethal Trifecta: Connected AI Accounts** - The real risk isn't just pasting data; it's connecting work accounts (Gmail, Slack, Outlook) to personal AI tools.
- 6 (10:35) **Why Bans Don't Work** - Companies like Samsung, JP Morgan, and Apple banned AI, but employees kept using it on personal devices.
- 7 (11:18) **The Three-Part Solution** - The host presents the actual fix: a sanctioned tool, a plain-language policy, and real training.
+ Full timestamped outline available in the app
Show Notes
Ever wondered what happens when your secret AI chats turn into courtroom exhibits? Today, we're diving into shocking stories of CEOs, engineers, and companies tripping over their own AI footprints—even after deleting chats. Spoiler: privacy and governance are more than just buzzwords—they're your new best friends.
KEY TOPICS
• Over 90% of companies’ employees secretly using AI tools for work
• How deleted chats can still haunt you in court (hello, Samsung's internal ban lift)
• The lurking dangers of connected AI accounts and prompt injections
• Real-life cases: a CEO’s illegal strategy, trade secrets leaked, and a major deal derailed
• The “lethal trifecta”: private data, outside content, and sent info at risk
• Why bans don’t work and what actually does (spoiler: good tools, clear policies, real training)
• A simple question to ask yourself before pasting into any AI tool
• The skyrocketing costs of uncontrolled AI use and why governance is your insurance
• The importance of proactive policies before a crisis hits
Chapters
00:00 - The shocking story of a CEO who deleted ChatGPT chats but still got sued
00:30 - Over 90% of companies' employees secretly using personal AI for work
01:00 - The dangers lurking in connected AI accounts and prompt injections
01:30 - Samsung's internal AI ban lift and what happened next
02:00 - Engineers uploading confidential code and internal meetings recorded by AI
02:45 - How even well-intentioned AI users can inadvertently leak secrets
03:15 - The legal case where deleted AI chats still surfaced in court
04:00 - The CEO’s illegal plans using AI to dodge a deal, and the evidence that betrayed him
05:00 - The hidden risks of connecting AI to other work tools like Gmail, Slack, and Outlook
05:40 - How AI-connected accounts can be hijacked and the "lethal trifecta"
06:00 - Why criminal hackers prefer employees with connected AI tools
06:30 - The flaws of bans and the 3-step solution for AI governance
07:00 - The simple question to ask before pasting into any AI tool
07:38 - How anonymized prompts can save your job and your reputation
08:00 - The shocking cost of unchecked AI breaches—over 9 million dollars on average
08:45 - Why governance beats strict bans—building the right policy and training
09:00 - How proactive AI management companies safeguard everyone
09:45 - Your quick checklist: tools, policies, and training
I wrote Moving Target because overconfidence is the enemy. Hardcover, paperback, Kindle, and