Summary

  • Adobe clarifies it won't train AI on customer data.
  • It says that consumers retain ownership over their content.
  • Adobe's updated terms of service, coming soon, will have clear and concise language.

Generative AI and other similar technologies are all the rage these days, but it's also important to navigate the nascent regulatory and privacy hurdles along the way. At its WWDC 2024 conference a few hours ago, Apple was especially keen on emphasizing the privacy aspects of its "Apple Intelligence" capabilities in a bid to win consumer trust. At the opposite end of the spectrum, Adobe found itself in hot water with customers after it was discovered that its updated Terms of Service (ToS) allowed the company to train its generative AI models on user content. The firm has now reversed course on the topic.

Adobe won't train its generative AI models on your data

Credit: Image generated by Simon Batt with Adobe Firefly Image 3

Adobe has been all-in on AI since the past few months, which is why eyebrows were raised when its legalese suggested that the company could train its AI models on customer data. However, in a new blog post, Adobe says that it won't engage in any such activity, nor will it claim any ownership of customer content. It says that it only accesses customer content when there are legal mandates to do so. That said, it will update the language in its ToS to better clarify its stance on customer-generated data.

What is Adobe's stance on ownership of consumer content?

Credit: Image generated by Simon Batt with Adobe Firefly Image 3

Adobe claims that its customers have complete ownership over their content, and while their data will not be used to train generative AI tools like Adobe Firefly, users can opt in to share usage data and other content characteristics to improve other technologies like masking and machine learning-powered background removal. Additionally, Adobe has emphasized that it does not scan data stored locally on your machine, but if you do upload data to its cloud platform, it will be automatically scanned to identify child sexual abuse material (CSAM).

The company has noted that with AI technologies evolving rapidly, it is crucial to have clear policies surrounding the topic. To that end, it will release its updated ToS next week with a focus on concise and relatively plain legalese that's easier to understand.