WhatsApp is defending the introduction of Meta AI to the App
Credit: Shutterstock: Herigki Tu
Generative AI has become a reality on WhatsApp, and the outcry has been swift as concerns over privacy and security have emerged.
However, in an attempt to defend the new Meta AI function and reassure its users, WhatsApp have highlighted that personal messages with friends and family are off limits. They have also confirmed that talking to an AI provided by Meta does not in any way link personal WhatsApp account information on Facebook, Instagram, or any other apps provided by Meta. While they accept that the ‘optional’ AI tool cannot be turned off, it can simply be ignored.
What does Meta AI do on WhatsApp?
They liken the ever-present distinctive Meta AI logo, with its blue circle with pink and green splashes, to other features with permanent options, such as ‘channels’ and ‘status’. In addressing its permanent presence, they have stated that Meta AI, through WhatsApp, is an optional service from Meta that can answer questions, teach something new or help users to come up with new ideas.
Further, they advise that only data sent to Meta may be used to provide accurate responses or to improve Meta’s AI models. To this end, they are ensuring that chats with AI are visually distinct so it’s clear these are different from personal messages and conversations.
Even so, they remind users to only send messages to Meta with information that they are happy for it to know. It also confirms that there is the option to delete interactions with Meta AI, and that it is trained to limit the possibility of information about people, like shared names, from showing up in other chats.
Meta AI on WhatsApp sparks strong reaction
However, this has done little to appease users who have aired their frustration at not being able to remove the AI functionality from the App. For example, Polly Hudson summed up the mood in a recent Guardian column: “There are five stages of grief, but only two stages of discovering the little Meta AI circle on your WhatsApp screen. Fear, then fury.”
Dr Kris Shrishak, an adviser on AI and privacy, was also highly critical and accused Meta of “exploiting its existing market” and “using people as test subjects for AI”.
“No one should be forced to use AI,” he told the BBC.
Whatsapp introduces Advanced Chat Privacy to address Meta AI concerns
In an effort to address these concerns, WhatsApp has announced that it is introducing a new feature, ‘Advanced Chat Privacy.’ This new setting will be available in chats and groups and is designed to prevent content from being taken outside of WhatsApp. By activating it on each individual chat, users can be confident that others are effectively blocked from exporting chats, auto-downloading media to their phone, and using messages for AI features.
Method used for training Meta AI model under investigation
However, how Meta trained its AI model in the first place has added to the controversy surrounding the functionality. An investigation by The Atlantic has revealed that Meta may have accessed millions of pirated books and research papers through LibGen (Library Genesis) to train its Llama AI, one of the large language models operated by Meta.
Should this be proven to be the case, it is likely to further undermine the trust and morally ethical behaviour that the success of AI integration depends upon. While the need for powerful models has driven rapid progress, it cannot come at the expense of solid legal and ethical practices.
Meanwhile, the Information Commissioner’s Office told the BBC it would “continue to monitor the adoption of Meta AI’s technology and use of personal data within WhatsApp”.
“Personal information fuels much of AI innovation, so people need to trust that organisations are using their information responsibly,” it said.
Over to you: Does it bother you to have the Meta AI function on your WhatsApp, or are you happy to just ignore it?
