How do you address business users’ concerns that their data might be used for training AI models?
Rada Vassil
6 replies
People in business who use AI assistants based on OpenAI (GPT, DALL-E etc.) and other models, are concerned that their data will be used for AI models’ training and potentially appearing on the web.
Businesses treat all their data as sensitive. This includes emails, personal information, product details, etc
How do you address these concerns?
Replies
Yuki@yuuuuk
To be honest this is a great point. I'm working on an app called smartrazor which allows YouTubers to edit their videos much faster. To accomplish this though and provide a great customer experience we are looking to train our models based off of the videos people upload, without attributing specific users to specific data. In your eyes, is this problematic (given we do not sell this data to third parties)?
Share
@rada_vassil1 thank you so much, this is genuinely very useful and we'll to a deep dive here particularly on the EU topic
@yukioyama Privacy and security can get quite complex, especially when dealing with business users. Since you're training your model with user-uploaded videos, even if they're anonymized, you may need to obtain explicit consent from your users, IMO.
Also, users are very sensitive about ensuring their data/videos are not available to other customers in any way.
Also, if any of your users are from the EU, they have data protection laws you need to comply with to prevent any legal issues.
I'm opposite you.
I actually advise executives to stay away from AI tools for the exact reason you say (and other reasons).
I know of a case where AI compromised intellectual property - very costly!
The only "free" information I'm willing to write for the executives reading this is that you should be extremely careful about where you allow AI to be used by both you and your employees. The downside costs are much bigger than people realize at the moment. But as the hype cycle in AI subsides, more people will see this.
Be careful!
@rada_vassil1 You nail it perfectly that companies should use more secure solutions where applicable while being careful about what data they are sharing.