Posted on in Company, Limecast by John Luther
This week, we were shocked to learn that tech companies are tricking hapless users into their AI data harvesting programs.
In other words, these companies will use your content to train AI models, generate content, and who knows what else unless you explicitly opt out. They assume the user's consent by default.
This is like someone robbing your house and saying, "Well, you didn't tell me I couldn't take all your stuff, so it's kinda, like, on you, man?"
Perhaps the most brazen offenders are UDemy and LinkedIn. Recently, they started automatically opting users into their generative AI training programs (UDemy enthused, "If you want to participate, no action is needed!"). UDemy went even further than LinkedIn and made it impossible for users to opt out of the program until next year.
This behavior is obnoxious. Sadly, it will become more prevalent as Big Tech desperately tries to convince investors that generative AI is not the biggest digital boondoggle since drkoop.com.
Sadder still, avoiding such nefarious tactics is yet another painful fact of life on the modern Internet. As the always droll New Yorker put it: "How to Opt Out of A.I. Online (Hint: You can't, really.)".
Nevertheless, we wanted to make a public vow about Limecast, our upcoming podcast product:
We will never collect customer or user data for any "AI" purpose.
In addition, we will actively block AI companies from scraping Limecast customer's content. We realize this will be tricky business, because these companies are very crafty and will flout even the most lawyered-up megatecholopoly's terms to steal content.
We have not decided yet what, if any, automation features we will build into Limecast. If we do, they will have tangible, practical value for customers and be enabled only through explicit "opt-in" actions. For example, machine-generated text transcripts and chapter markers would be pretty handy; generating terrible podcasts with data obtained through nefarious means, not so much.