AI training: Renowned US websites block Apple bot

Apple is currently intensifying its efforts in the field of artificial intelligence. However, many large US websites are now apparently putting obstacles in the way of the iPhone company and blocking the AI ​​training of the Apple bot.

Just in June, Apple presented its new AI system, Apple Intelligence, at its developer conference WWDC. According to the US company, this system is intended to serve as “the personal intelligent system for iPhone, iPad and Mac”.

But AI training has been a bit bumpy for Apple so far. Because Wired reported, numerous US websites are blocking Apple’s AI scraping.

AI training: Apple is not allowed to train

The Apple Bot has been around since 2015 and was initially used to improve Apple products such as Siri or Spotlight. But now Apple has introduced the Apple Bot Extended, which comes with extended functions. The data that this AI bot collects can also be used to train AI models.

At its developer conference, Apple explicitly pointed out that it would offer website operators special options. This is intended to ensure that only data from websites that have agreed to this is included in AI training.

But according to the WiredAccording to the report, it is mainly the big players who are blocking this, including the New York Times and also Facebook and Instagram. But Tumblr, the Financial Times, the USA Today network and Wired's parent company Condé Nast are also making use of their exclusion from Apple's AI training.

Concerns about intellectual property are at the heart of this. This is why publishers, for example, can block the Apple Bot Extended. All they have to do is update the “Robots Exclusion Protocol” text file on their websites.

Media companies in particular are blocking the Apple bot

Analyses show that the percentage excluding Apple is still relatively small. This is because news and media companies in particular are excluding the new Apple bot from their websites.

The Canadian startup Originality AI examined a sample of 1,000 heavily frequented websites. Of these, only around seven percent blocked the Apple Bot Extended. A similar analysis by the AI ​​agent monitoring service Dark Visitors came to around six percent.

Taken together, these efforts suggest that the vast majority of website owners either do not object to Apple's AI training practices or are simply unaware of the ability to block Applebot-Extended.

However, the percentage is significantly higher for news websites, as an analysis by data journalist Ben Welsh shows. Around a quarter of the 1,167 mainly English-language and US-based sites examined block the Apple Bot Extended.

By comparison, 53 percent of news websites block OpenAI's bot, while Google's bot, called Google-Extended, is blocked by almost 43 percent of these websites. Welsh explains to Wiredthat the number has “gradually moved upwards” since he began observing it.

Expert Welsh cannot explain why some publishers allow AI bots and others do not. He suspects that some companies have entered into licensing agreements and are paid to allow the bots.

Also interesting: