WhatsApp launches world's first 'incognito' mode but there's a catch
Privacy is something that many people value in their messaging apps, as while you might not even be saying anything that would be damaging or embarrassing if exposed, it's natural to want your conversations to be secure.
It's why so many people have gravitated towards apps like Signal and WhatsApp over the years, as end-to-end encryption alongside additional cybersecurity protections ensure that your chats are read and received only by those involved.
There have been newly proposed apps that take advantage of alternative forms of technology to ensure an even more secure service, yet most people will opt for the convenient yet reliable option — especially as that's what their friends and family tend to use.
One newly added aspect of WhatsApp in particular has remained a privacy concern for some, however, as conversations with Meta's in-built AI chatbot weren't perhaps as secure as users might have hoped.
Advert
That's changed now, as reported by the Metro, as WhatsApp has officially added an 'incognito mode' to conversations made with the Meta AI tool, ensuring that your chats are only accessible to the user.
Anything you say to the Meta AI chatbot won't be visible to the tech company behind it, and all messages sent while in incognito mode disappear after a predetermined period of time and leave no trace of ever existing.
This has prompted some, however, to raise concerns surrounding the safety of the feature — especially considering the alarming number of people already discussing complex mental health topics and suicide with other chatbots.
Speaking to the Metro, Andy Burrows, chief executive of online safety charity Molly Rose Foundation, proclaimed that "Meta's decision to roll out encrypted, disappearing AI conversations is high-risk and deeply irresponsible move," clarifying that it would make it 'impossible' to find evidence in situations where the chatbot could have contributed to the harm of a user.
Meta have clarified that the chatbot built into WhatsApp has a number of safety features that prevent the model from discussing or responding to harmful topics, instead offering 'helpful information' in situations where it applies.
There remains ongoing conversations surrounding the safety of AI tools for people suffering from mental health emergencies though, and the frequency of hallucinations and instances of sycophancy only add to the fear that the right decision won't always be made.
Without the ability to access chats, Meta hypothetically has no way of knowing what might have potentially gone wrong in instances of potential harm, making preventative measures and accountability challenging.