Slack trains machine learning models on user messages, files, and other content without explicit permission. The training is opt-out, meaning your private data will be leaked by default. To make matters worse, you'll have to ask your organization's Slack administrator (HR, IT, etc.) to send an email to the company asking them to stop. (You can't do it yourself.) Welcome to the dark side of the new ai training data gold rush.
x.com/QuinnyPig” rel=”nofollow noopener” target=”_blank” data-ylk=”slk:Corey Quinn;elm:context_link;elmt:doNotAffiliate;cpos:2;pos:1;itc:0;sec:content-canvas”>Corey QuinnDuckBill Group executive, x.com/SlackHQ/status/1791277972722188311″ rel=”nofollow noopener” target=”_blank” data-ylk=”slk:spotted;elm:context_link;elmt:doNotAffiliate;cpos:3;pos:1;itc:0;sec:content-canvas”>stained politics in propaganda Slack Privacy Principles and published about it in x (ai-powered-features-on-user-messages-files” rel=”nofollow noopener” target=”_blank” data-ylk=”slk:via;elm:context_link;elmt:doNotAffiliate;cpos:5;pos:1;itc:0;sec:content-canvas”>through PCMag). The section says (emphasis added), “To develop ai/ML modelsour systems analyze customer data (e.g messages, content and files) sent to Slack as well as Other information (including usage information) as defined in our Privacy Policy and your customer agreement.”
In response to concerns about this practice, Slack published a ai” rel=”nofollow noopener” target=”_blank” data-ylk=”slk:blog post;cpos:6;pos:1;elm:context_link;itc:0;sec:content-canvas” class=”link “> on Friday night to clarify how its customers' data is used. According to the company, customer data is not used to train any of Slack's generative ai products, for which it relies on third-party LLM, but is instead fed into its machine learning models for products “such as customer recommendations.” channels and emojis and search results.” For those apps, the post says: “Slack's traditional machine learning models use de-identified aggregate data and do not access message content in direct messages, private channels, or public channels.”
A Salesforce spokesperson reiterated this in a statement to Engadget, also saying that “we do not build or train these models in such a way that they can learn, memorize, or reproduce customer data.”
<div class="twitter-tweet-wrapper” data-embed-anchor=”2b87d65b-5f23-51d8-9b09-9d9878acf246″><blockquote placeholder="" data-theme="light" class="twitter-tweet”>
Sorry Slack, are you doing what the fuck with direct messages, messages, files, etc. of the users? I'm sure I'm not reading this correctly. twitter.com/6ORZNS2RxC;elm:context_link;itc:0;sec:content-canvas” class=”link “>pic.twitter.com/6ORZNS2RxC
-Corey Quinn (@QuinnyPig) twitter.com/QuinnyPig/status/1791220276350390575″ rel=”nofollow noopener” target=”_blank” data-ylk=”slk:May 16, 2024;elm:context_link;itc:0;sec:content-canvas” class=”link “>May 16, 2024
The opt-out process requires you to do all the work to protect your data. According to the privacy notice, “To opt out, ask your organization or workspace owners or primary owner to contact our customer experience team at [email protected] with the URL of their space of work/organization and the subject 'Request to opt out of the global Slack model'. .' We will process your request and respond once the opt-out has been completed.”
The company x.com/SlackHQ/status/1791277972722188311″ rel=”nofollow noopener” target=”_blank” data-ylk=”slk:replied;elm:context_link;elmt:doNotAffiliate;cpos:7;pos:1;itc:0;sec:content-canvas”>answered to Quinn's message on x: “To clarify, Slack has platform-level machine learning models for things like channel and emoji recommendations and search results. And yes, customers can exclude their data to help train those (non-generative) machine learning models.
It's unclear how long ago the Salesforce-owned company introduced the detail into its terms. It is misleading at best to say that customers can opt out when “customers” does not include employees who work within an organization. They need to ask whoever handles Slack access at their business to do it, and I hope they do.
Inconsistencies in Slack's privacy policies add to the confusion. One section says: “When developing Al/ML models or analyzing customer data, Slack cannot access the underlying content. “We have several technical measures to prevent this from happening.” However, the machine learning model training policy apparently contradicts this claim, leaving much room for confusion.
Additionally, the Slack website ai” rel=”nofollow noopener” target=”_blank” data-ylk=”slk:marketing;elm:context_link;elmt:doNotAffiliate;cpos:9;pos:1;itc:0;sec:content-canvas”>marketing its premium generative ai tools say: “Work without worries. Your data is your data. We don't use it to train Slack ai. “Everything runs on Slack’s secure infrastructure and meets the same compliance standards as Slack itself.”
In this case, the company is talking about its premium generative ai tools, separate from the machine learning models it trains on without explicit permission. However, as PCMag points out, implying that all your data is safe from ai training is, at best, a very misleading statement when the company can apparently choose which ai models that statement covers.
Update, May 18, 2024, 3:24 pm ET: This story has been updated to include additional information from Slack, which posted a ai” rel=”nofollow noopener” target=”_blank” data-ylk=”slk:blog post;elm:context_link;elmt:doNotAffiliate;cpos:12;pos:1;itc:0;sec:content-canvas”>blog post explaining their practices in response to community concerns.
<script async src="//platform.twitter.com/widgets.js” charset=”utf-8″>
