AI models are coming for all of your data, personal and business
posted Sunday May 19, 2024 by Scott Ertz
When AI technology first hit the mainstream a couple of years ago, the biggest concern came because of where the data came from. It appeared that the models were being trained on data that the companies behind them didn't have permission to train off of. They were scraping websites, blogs, newspapers, artist portfolios, music libraries, and more to train their data. While that was a violation of copyrights in many cases, the next generation of AI is going to find its data in new and even scarier places - your devices.
Slack may be spying on chats
Following the launch of Slack AI earlier this year, customers have been pouring through the company's data policies looking for how the system was trained. The policies created nothing but confusion for users, leading to discussions on social media about how to deal with the policies. The general consensus was, that because the policies were not clear, users should opt out of any data usage. This might mean that search would stop working on your account, but it would also prevent training on your data.
One such conversation on Threads (I know, I was surprised, too) was responded to by an engineer at Slack. He pointed to a blog post from the company that states that they do not use customer data for training. However, customers pointed to various other documents from the company that say that they absolutely can use that data, even if they currently are not. Slack's privacy principles state,
Machine Learning (ML) and Artificial Intelligence (AI) are useful tools that we use in limited ways to enhance our product mission. To develop AI/ML models, our systems analyze Customer Data (e.g. messages, content, and files) submitted to Slack as well as other information (including usage information) as defined in our privacy policy and in your customer agreement.
However, the Slack AI page states,
Work without worry. Your data is your data. We don't use it to train Slack AI.
While technically both can be true at the same time, the inconsistency creates a scenario in which the company can state today that they do not use your data for training, but change that policy tomorrow without any warning, because their policies state explicitly that they can use it. Customers are urging the company to change the policy to be clear and consistent, or will urge people to change services to someone who is more clear about their data usage.
Google will be spying on your calls
Following this controversy, Google announced this week during Google I/O, that they have a new AI-powered feature coming to your phone app. The system will be able to listen in on your calls and help determine if the call you are receiving is a scam call.
Initially, you might think, "This could help my parents avoid the tax scams or the gift card scams," and that is perfectly reasonable. The concept is a nice one - protect people from themselves and their human nature. However, the system requires that Google listen to every phone call, translate the call into text, and compare it against an ever-growing database of questionable topics. That means that Google will be listening to everything that you and the other people on a call are saying.
Let's say you're a software engineer and you are discussing a new product you are working on with colleagues. You might not want Google to be listening to that conversation, as they are technically a competitor. But, more importantly, do you really want your conversations with friends and family logged away? Of course not.
Conclusion
As AI infiltrates more aspects of our lives, we need to be cautious about how it works and where the data is coming from. Many companies are making a lot of money off of the data of other people, and you could be feeding into that system without even knowing it.
