Three months ago, Apple debuted a tool that makes it easy for companies to opt out of its AI training, and it appears that several major companies have used it.
Unlike Google, Apple has been offering companies millions of dollars to allow its AI to use crawlers to scrape data to train Apple Intelligence. As part of the deal, companies can elect for certain files to be omitted from the AI's training. This omission is controlled by an extension to Applebots. Applebots were originally released in 2015 and were designed to crawl the internet to power features like Siri and Spotlight, however, they have since been repurposed to train Apple Intelligence.
Apple has included a new extension, known as Applebot-extended, that lets website owners tell Apple not to use their data. In a recent report, Wired lists several significant companies that have been elected out of the training. These include The New York Times, Facebook, Instagram, Craigslist, Tumblr, Financial Times, The Atlantic, USE Today and Conde Nast.
According to Wired, two main studies have shown that around 6% to 7% of high-traffic websites are blocking Applebot, proving that companies either don't mind Apple's training or are unaware of the option to refuse it. Interestingly, another analysis by journalist Ben Welsh found that around a quarter of websites blocked Apple, compared to 53% of companies blocking Open AI's bot and Google AI's Google Extended is blocked by around 43%.
Companies crawling sites for AI training has been a point of contention recently, with Apple having been accused of using stolen YouTube captions to train its AI. However, Apple has made it clear that any data taken from the pile, the name given to the collected data, was not used to train Apple Intelligence. It was also recently revealed that Apple originally trained its AI on Google hardware, as listed in a recent research paper.
AI is growing rapidly at the moment, and how companies gather their data will be under close scrutiny. Apple, and other companies, offering companies an out is important, especially as certain sites contain profiles and images from the general public.
For more information about Apple Intelligence, and what it brings, you can check out our full breakdown of Apple's WWDC presentation.
More from Tom's Guide
- Grok gets an upgrade in iOS — make and analyze images in X with AI
- Move over Midjourney — Freepik Mystic is the most realistic AI image generator I've tried
- I used ChatGPT Advanced Voice to go on a time-travel adventure — and you have to hear it
