Apple Introduces Hate Speech Detection In Next iOS Update

Published 2024-10-1 by NAITO Yoru

Apple has recently announced via a website post that detection of hate speech will be included as a feature of the next iOS update. Apple will be harnessing the power of its new Apple Intelligence to power this revolutionary new technology.

Detailed in the post is the reasoning for developing this new feature. The post states "While Apple has always been a privacy-focused company bringing customers the benefits of privacy, security, and ease of use, we have noticed the potential pitfalls of providing these benefits. One of those pitfalls that we have decided to address is that of hate speech. With the proliferation of hate speech and misinformation, we have noticed a rise in bigotry across the internet. We have noticed that while the increasing of moderation in mainstream social media and content platforms has been very effective in halting the spread of hate speech, many people who are affected by this rise in moderation simply move to other platforms where they can continue spreading hate and bigotry. This problem is worsened by the fact that many of these dubbed 'alt-tech' platforms create echo chambers that further reinforce these bigoted beliefs. We at Apple have come up with a solution to hopefully mitigate this."

The rest of the blog post goes into technical detail about how this hate speech detection will be conducted. The technically savvy can go there for all the finer details but we will be explaining in more simpler terms how this will occur.

With the introduction of Apple Intelligence comes a wide variety of opportunity for privacy-respecting surveillance. The new AI will actively scan your device in real time and detect anything that could be considered hate speech. Because your device's content and screen content can be scanned by this AI in its entirety, it will be able to distinguish context like whether the detected hate speech was just part of a movie or if someone in a private chat said something hateful. If Apple's AI detects that the user of the device is the one either saying hate speech or actively engaging with others who spread hate in a positive way, the AI will signal via your Identifier for Advertisers (IDFA) to tell advertisers to give you ads in apps and on the internet resources and information about hate speech, its effects, and where you can learn to stop hate. The device may also display a pop-up with resources to learn about the effects of hate speech and how to prevent it. Apple has also announced that they will be updating their Terms of Service with a clause stating that continuance of spreading and/or positively engaging with hate speech after repeated pop-ups may result in a termination of your Apple account.

This new feature will also have the capability to provide details about specific instances of hate speech detected and explain within context why it is wrong. Apple hopes that by educating people on the truth behind instances of hate speech, people will learn the harmful effects it has on society and the falsities behind it. Apple affirms that due to Apple Intelligence's ability to distinguish contexts, this feature will not be able to be used the same way as user-made detection of hate speech and will have no bearing on account status as long as the device user isn't spreading or positively interacting with hate speech.

This announcement has been met with some backlash from privacy advocate groups stating concerns that this is unnecessary surveillance and infringes upon the privacy rights of its users. Apple has responded assuring that this new feature does respect the privacy of its users. Apple confirms that all device scanning will be done on-device and AI scanning of your device will never be done in the cloud. However, if hate speech is repeatedly detected, the updated ToS states that Apple reserves the right to have the infringing data sent to Apple to be reviewed if the AI is absolutely certain that repeat violations have been made. Apple has stated that during testing, the chances of a false positive from hate speech detection being made has less than a 1 in 1 trillion chance per year of happening.

With the problem of hate speech worsening across the internet and the world, this brings a breath of fresh air to those fed up with hate. Hopefully Apple's new privacy-respecting device scanning and surveillance will bring us at least one step closer to stopping hate speech for good.