Google has announced new updates to its terms of service, aiming to provide increased protection for users against two specific types of lawsuits. The first scenario involves users being sued for using data to train artificial intelligence models without permission. This covers instances where Google’s own models are being sued for training with the LLM algorithm without authorization. However, there may be cases where the plaintiffs choose to sue the users of these models instead. In such cases, Google will take responsibility for any damages incurred.
The second scenario pertains to the creation of outputs. Google considers the owner of the work to be the customer who inputs their text into the AI model. If the generated outputs violate someone else’s copyright, Google will assume liability, provided that the input was not intentionally designed to infringe on others’ work.
These measures are aimed at safeguarding users and instilling confidence, similar to the approach announced by Microsoft in September.
TLDR: Google has updated its terms of service to protect users from being sued for training AI models without authorization and for generating outputs that violate copyright. They will assume responsibility in these cases to ensure user confidence.