Key Takeaways
- Starlink updated its Global Privacy Policy in January, allowing the use of customers’ personal information to train artificial intelligence and machine learning models.
- The updated policy extends the use of data to the company’s affiliates, service providers, and third-party collaborators, who may also use it to train AI for their own purposes.
- Starlink has yet to provide more details on what data subsets specifically are being used to train the models.
- Customers are opted in by default but can opt out by following the instructions on the Starlink website.
What Happened
Starlink, a satellite network operated by the space exploration company SpaceX, delivers high-speed internet to millions of people around the world and is aiming to provide global internet connectivity. SpaceX founder Elon Musk launched the project in 2019, and it has since surged in popularity, struggling to meet the demand of a growing waitlist. In January, Starlink updated its Global Privacy Policy, which now states that personal information can be used “to train [their] machine learning or artificial intelligence models” and that the company may also share this data with “affiliates, service providers, and third-party collaborators” who can also use it to train AI, “including for their own independent purposes.”
Privacy and Governance Concerns
The data that Starlink collects on its users includes contact details (name, phone number, email address), credit card information, and location data. The policy also states that they may collect “communication information such as audio, electronic, or visual information,” and “inferences we may make from other personal information.” Although the amount and type of data being collected haven’t changed, this policy update extends its use to service providers in the name of creating more robust AI models. Customers are opted in by default but can opt out by following the step-by-step instructions provided on the Starlink website.
Starlink has yet to provide more details on exactly which data subsets are used to train the models, or whether the data will be anonymized, aggregated, or retained in raw form. This ambiguity is making some users apprehensive, with one Georgetown professor, Anupam Chander, commenting on how “there’s often perfectly legitimate uses of your data, but there is no clear limit on what kind of uses it will be put to”. Privacy advocates and consumer rights groups have highlighted that harvesting personal data for AI training may introduce new risks and could create additional avenues for misuse.
Why It Matters / Policy Considerations
Overall, the lack of detail about which data subsets are being used and in what form points to a broader lack of transparency and accountability for Starlink users. It raises concerns regarding how that data could be used beyond Starlink’s control, potentially exposing users to surveillance or profiling risks.
