Slack is one of the most popular platforms for business communication, along with Microsoft Teams. It has recently implemented a policy of using customers’ data to train artificial intelligence models.

This data includes not only conversations on company-facing channels, but also your private messages and files. This is done by default and without asking for explicit permission.

Because this is important. This case highlights the privacy risk posed by more companies developing AI tools using large amounts of user data, often in a non-transparent manner. This also raises questions about the level of control people and companies have over their information in cloud services.

Made by Slack. On a confidential page on its website about its privacy policies, Slack explained that its systems “analyze customer data (like messages and files) as well as other information (like usage analytics) sent to Slack to train new service models.” Artificial intelligence, by default, and without asking special permission.

The only way to exclude a company’s data is for a manager on your team in Slack to send an email to the service requesting one. to give up spherical. Individual users cannot perform this operation.

privacy concerns. Slack says it has technical controls in place to block access to certain forms of content and that data won’t leak between workspaces, but says feeding private chats and sensitive files into AI systems poses a risk.

Another way to request an exclusion is to have to send an email from the administrator, rather than it being a simple and integrated process (and not executed by default). dark pattern from those we have known for a long time. It’s a way to prevent companies from seeking exclusion.

Slack’s argument. In a reassuring statement, they said they use these techniques for “things like channel recommendations, emojis, or search results.” It’s worrying that they’re using private conversations and sensitive files by default for something as trivial as suggesting emojis in reactions.

They assure that the data is “anonymised and aggregated” and that customers can request exclusion, but there is a risk of information leakage when working with private data.

lingering doubts. With usage focused on minute details like suggesting emojis or sorting calls, many people wonder why Slack needs access to private conversations for something like this.

The promise that data will not leave your infrastructure does not eliminate the risk of breach or misuse. And this case is an example of the privacy dilemma posed by the rise of artificial intelligence, as well as the temptation for companies to have to feed their models with sensitive user data and opaque techniques.

in Xataka | There’s a superpower that boosts your productivity almost instantly: saying “not now”

Featured image | Stephen Phillips (Host Reviews) on Unsplash