Apple’s new AI features for iOS 18, iPadOS 18, and macOS 15 are set to be announced during the WWDC 2024 keynote. While we’ve learned a lot about what Cupertino might unveil, a new report shows that it might be the only major tech company to train its AI legally.
With the rise of OpenAI’s ChatGPT, Microsoft’s Copilot, and Google’s Gemini, we have also reported that these AIs are being trained illegally. For example, in late 2023, The New York Times sued OpenAI and Microsoft for copyright infringement.
On the other hand, Apple has been reportedly trying to license work from Conde Nast, IAC, and NBC News to train its AI. Whether the deal went through is unclear. Still, since Apple’s AI reportedly offers on-device functionality, it means Apple won’t be connecting to the internet, so it won’t have copyright issues.
For that matter, Apple is probably making a deal with Google or OpenAI for more complex tasks. With that, any copyright issue would be on these companies but not Apple.
According to AppleInsider, people familiar with the matter told the publication that Apple’s Large Language Model “preserves privacy because it does not require an internet connection for basic text analysis.”
Still, sources familiar with Apple’s AI test environments told the publication that there were “seemingly little to no restrictions to prevent someone from using copyrighted material in the input for on-device test environments.”
With that, it’s possible that Apple is trying to avoid word-for-word reproduction of copyrighted material, but we’ll have the company announce its AI plans to prevent copyright-violating training.
As mentioned by AppleInsider, Cupertino may be focusing its efforts on basic tasks that don’t require the internet (such as creating Apple Music playlists and auto-summarizing text on Pages) while letting complex texts for its partners, as they’ll be responsible for the outsourced material.
Below, you can learn more about Apple’s AI plans.
