
Your work laptop may be training the AI that eventually replaces you — and Meta just gave us a preview of how creepy that future could get.
In this episode of Legitimate Cybersecurity, Frank Downs and Dr. Dustin Brewer discuss reports about Meta’s Model Capability Initiative, a program designed to collect employee computer activity such as mouse movements, clicks, keystrokes, and occasional screenshots to help train AI systems.
This is not just another “employee monitoring” story. Traditional workplace monitoring was usually framed around security, compliance, or productivity. This is different: the purpose is AI training. That means your daily work behavior could become the dataset that helps AI agents learn how to navigate real software, real workflows, and real human messiness.
Frank and Dustin break down why AI agents still struggle with ordinary computer tasks, why companies are desperate for human workflow data, and why “trust us” is not a privacy policy. They also discuss the reported data exposure concerns, the difference between “no indication of improper access” and actual assurance, and what this means for anyone using a corporate device for personal life.
Main discussion threads:
* Why Meta wanted real employee computer activity to train AI
* How AI workplace monitoring differs from traditional enterprise monitoring
* Why AI agents still struggle with menus, clicks, and normal workflows
* The privacy problem with keystrokes, screenshots, prompts, and performance data
* Why “no evidence of misuse” does not mean “nothing happened”
* Whether this becomes normal as compute gets cheaper
* Why you should stop using your work laptop for personal activity
The big warning: if your employer owns the laptop, assume the laptop is not yours. And now, assume AI may be watching too.
Media/interview: mailto:[email protected]
Audio: https://legitimatecybersecurity.podbean.com/
Chapters:
00:00 Is Meta showing us the future of work?
01:36 What Meta reportedly tracked
02:16 Normal monitoring vs AI training
04:50 Why AI agents need human workflow data
08:04 Is Meta building AI for users — or for extraction?
09:31 Employees, trust, and the 30-minute opt-out problem
15:36 The data exposure issue
17:17 “No indication of improper access” is not enough
20:15 Is workplace AI surveillance here to stay?
23:42 Stop using your work laptop for personal stuff
26:31 Final warning: AI is watching
#AI #Cybersecurity #Meta #EmployeeMonitoring #WorkplaceSurveillance #ArtificialIntelligence #Privacy #DataPrivacy #AIAgents #FutureOfWork #TechNews #LegitimateCybersecurity
Otros episodios de "Legitimate Cybersecurity Podcasts"



No te pierdas ningún episodio de “Legitimate Cybersecurity Podcasts”. Síguelo en la aplicación gratuita de GetPodcast.








