le: All online content should be accessible for AI training, unless publishers opt-out //

Google wants AI to have access to all digital content, unless publishers opt-out.

This proposal was made by the search engine in a submission sent to the Australian Government, urging policymakers to amend current copyright laws.

Why do we care? Should copyright laws change, it will be up to brands and publishers if they want AI to stop mining their content. They run the risk that rivals could generate very similar content without them being able do anything. This could be damaging to their brand identity and reputation.

What’s happening? Google sent a letter to the Australian Government stating that:

• “Copyright Systems that allow appropriate and fair usage of copyrighted contents to enable the Australian training of AI models on a wide range of diverse data while supporting working opt-outs to entities who prefer that their data is not trained using AI systems.”

Google has, of course, made similar arguments to the Australian Government before. It argued that AI should be allowed fair access to online content as a training tool. This is the first instance that Google has proposed an opt-out as a way to address previous concerns.

What would happen? Google has not yet developed a plan, but it has said it is interested in holding discussions on a web standard created by the community that functions in a manner similar to the robots.txt, which allows publishers to opt-out of search engines crawling content.

What did Google say? Danielle Romain, vice president of trust for Google Search, spoke on this subject in a recent press release. She said:

• “We think everyone can benefit from a dynamic content ecosystem. Web publishers must have control and choice over their content and the ability to gain value by participating in the web eco-system. We acknowledge that the existing controls for web publishers were created before new AI use cases and research.
• As new technologies are developed, the web community has the opportunity to develop standards and protocols which will support the future development of the web. robots.txt is one such web standard that was developed by the community nearly 30 years ago. It has been a transparent and simple way for web publishers and web developers to control how search engine crawlers read their content. “We believe that it is time for the web community and AI to explore other machine-readable ways for web publishers to choose and control emerging AI and research uses cases.”

Deep dive. For more information, read Google’s announcement ‘Evolving Choice and Control for Web Content.’

The article Google says that all online content should be made available for AI training, unless publishers opt-out first appeared on Search Engine land.