Slack trains machine learning models to read user messages, files, and other content without obtaining explicit consent.
The platform operates by default using your personal data, and you must unsubscribe from training by requesting it from your HR administrator or IT department in your organization by sending an email to the company to request cessation, and you cannot do it yourself.
Kory Quinn, one of the top executives at DuckBill Group, discovered this policy in Slack’s privacy principles and published it via X.
The policy states that in order to develop artificial intelligence and machine learning, customer data submitted to Slack, including messages, content, files, and usage information, is analyzed according to the privacy policy and customer agreement.
According to the privacy policy, unsubscribing requires a request from the owners of your institution or workspace or the primary owner to contact Slack’s customer experience team using the workspace or institution’s address. The request will be processed and responded to once the unsubscribe process is completed.
The company responded to Quinn’s tweet saying, “Slack owns platform-level machine learning models for several things like channel recommendations, emojis, and search results. Customers can exclude their data from assisting in training machine learning models”.
It is unclear when Salesforce’s partner company updated its privacy policy.
The increasing contradictions in Slack’s privacy policies lead to confusion, as one section indicates that Slack cannot access core content during the development of machine learning models or analyze customer data in another way due to technical measures preventing that.
The strategy of training artificial intelligence models seems to contradict this section, causing significant confusion.
The marketing page for Generative AI Tools states: “We do not use your data to train artificial intelligence models. Everything is done through the secure platform infrastructure, in compliance with the standards enforced by Slack itself”.
In this context, the company refers to distinctive tools for generative artificial intelligence that operate independently of machine learning models and train on data without explicit permission.
Assurances are provided that all your information remains safe from the use of artificial intelligence as a means of espionage, while the company defines the artificial intelligence models that may include this information.
