Major Security Flaw In Rabbit R1 Code Could Have Allowed Hackers To Access Customer Data, Brick All Devices

The Rabbit R1 AI device launched a few months ago, and since then, its existence has been plagued by controversies, one after the other. We reviewed the hardware and gave it an abysmal 2/10 rating, citing issues like half-baked technical implementation of several features, not being able to provide enough value to complement a smartphone, connectivity issues, and more. The internet echoed our thoughts, also discovering that Rabbit used to be a shady NFT company and that the R1 could just be used as a smartphone app without requiring dedicated – and rather expensive – hardware.

We also emphasized how releasing half-baked tech after over-promising is the worst tech trend of 2024. And now, to put icing on the cake, a group of reverse engineers under the name Rabbitude has discovered a major security flaw in the R1’s source code.

What actually is Rabbitude?

Rabbitude describes itself as a reverse-engineering project that hacks and experiments with the R1, and then reports its findings publicly. Its members comprise R1 users and other people from the community who are interested in modding and jailbreaking the AI technology, so that it can be further improved. These activities allow the integration of various new features and functionalities, including a 24-hour clock, spoofing your GPS location, enabling touch screen, running apps like YouTube, and more. The group publishes its findings publicly so that other researchers and developers can further build upon their work, with the ultimate goal being to offer a better experience for all R1 users.

What security flaw was discovered by Rabbitude?

In a blog post, Rabbitude says that on May 16, it managed to gain access to the R1’s codebase, where it discovered that several API keys had been hardcoded. These keys could be used to read all responses ever given by the R1 (including those containing personal data), alter all R1 responses, and replace the R1’s voice. In fact, they could even be used to delete R1 voices, which would crash the backend and brick all devices.

In its initial investigation, Rabbitude noted that four API keys are exposed and are leveraged for services ElevenLabs, Azure, Yelp, and Google Maps. As a scathing parting note, the group stated that the R1 development team is aware of this flaw but has decided to ignore its feedback. As such, Rabbitude will not be publishing further details as it respects the privacy of R1 users, even if it does not respect the company itself.

Soon after the publication of this blog, Rabbit published a security update on its website, saying that it has now rotated the keys during a brief downtime. In its own ongoing investigation, it claimed that no critical systems or customer data has been accessed.

However, Rabbitude then published a follow-up blog post, criticizing Rabbit’s poor security practices. It noted that it had deliberately missed out on mentioning a fifth API key exposed in the source code, that is used to access Sendgrid. The group claims that the API key for this service is still active and can be used to access all emails sent on the r1.rabbit.tech subdomain, yet another breach of customer privacy. In fact, it can even be used to send emails to users from rabbit.tech email address.

So far, Rabbit has not commented on this revelation, but it is clear that this is a massive PR disaster for the struggling company, and one which it may never be able to recover from. Winning and then maintaining customer trust is one of the core principles for companies which deal with consumer data, and it turns out that Rabbit has failed on this front too.