AI Pioneer Warns: AI Wants Your Private Files

AI companies are running out of easy data — and the next target may be your private files, calendars, medical records, photos, and desktop activity. AI pioneer Dr. Jonathan Schaeffer joins Frank Downs and Dustin Brewer to explain why today’s AI tools are powerful, flawed, and increasingly hungry for personal data. In this episode of Legitimate Cybersecurity, Frank and Dustin talk with Dr. Jonathan Schaeffer, University of Alberta Professor Emeritus, AI pioneer, AAAI Fellow, entrepreneur, and founder of Synsara. They discuss why today’s chatbot boom is not the AI future many researchers imagined, why “hallucination” is the wrong word for AI errors, how AI companies depend on more and more data, and why desktop AI tools may create a new privacy boundary problem. The conversation also covers AI bias, manipulation, private data, local AI, regulation, data centers, environmental costs, and why solving AI’s safety and privacy problems should matter before the race to AGI gets even faster. Dr. Schaeffer’s key warning is that current AI systems do not understand the consequences of their answers, yet people increasingly treat them like trusted authorities. Media/interview: [email protected] Audio: https://legitimatecybersecurity.podbean.com/ Chapters: 00:00 — AI’s privacy problem is getting bigger 01:27 — Jonathan Schaeffer’s AI origin story 03:29 — Beating humans at checkers before Deep Blue 05:48 — Why modern AI feels like the wrong future 07:50 — Why “hallucination” is the wrong word 09:01 — How “chat” created false trust 10:32 — AI does not understand consequences 13:52 — Why AI companies are desperate for data 15:12 — Your private files are the real gold mine 16:32 — The hidden cost of “free” AI tools 20:44 — AI wants access to your desktop 22:50 — The safety, security, and privacy problem 24:05 — The AGI race is moving faster than safeguards 27:07 — Why Jonathan built private local AI tools 30:59 — The security risk nobody talks about 32:31 — Why AI systems need audits 34:21 — When AI answers become manipulation 39:13 — Influence, rage content, and algorithmic persuasion 42:21 — Why AI regulation cannot keep up 46:05 — Canada’s failed attempt to regulate AI 50:40 — Is it already too late? 55:16 — What polar exploration teaches us about AI risk 59:39 — Data centers, power, water, and responsibility 1:03:18 — Jonathan’s life advice: fun beats money #ArtificialIntelligence #AIPrivacy #Cybersecurity #DataPrivacy #ChatGPT #AISafety #Privacy #TechPolicy #LegitimateCybersecurity #Synsara