We now live in a world where almost everything we touch or do can be documented electronically.
We made the design decision to make mote content accessible on the open internet, but only accessible from short random URLs so that the probability of random discovery was very low (currently, about 150 in 1 septillion. If you're wondering, there are 24 zeros in a septillion).
We did this so that users wishing to access mote content - without having the mote extension installed - could do so without needing to create an account. We are looking at alternative approaches, and you can expect us to provide other options in the future. But - regardless of where the files are stored - there's still a fundamental trade-off between accessibility, on the one hand, and security, on the other.
Your mote audio files are played in the browser, and can be downloaded by you, or by anyone else with access to the page. We understand that this may be uncomfortable for some people, but let us briefly explain our thinking:
- If files couldn't be downloaded, a malicious actor could simply record a copy of the audio using a cell phone, or more sophisticated software capturing audio / video from the browser.
- We think it's best for all of us to assume that *any* content that we create online can be discovered, and potentially shared, beyond its intended recipient(s).
- If you need to communicate confidentially with a colleague or student, we would recommend exploring alternative tools.
You'll also note that the web pages hosting your voice notes include a headline title such as "Will Jackson left you a voice note...". We think this attribution is important. However, for users under age 18, we disable this titling, and their voice notes are therefore publically anonymous.
Finally, we've heard some people concerned that their voice could be spoofed. 'Deep fakes' are a real threat, especially for public figures'. However there's several reasons to feel reassured about this threat.
- Training a machine model to accurately replicate a voice still requires hours of audio content. Many of the more famous examples you may have seen are created mixing samples of pre-recorded audio with a machine model powered by hundreds of hours of speeches by, e.g., Barack Obama.
- Since the URL for every mote page is randomly generated, it would be hard (though in truth not technically impossible) for a malicious actor to discover all of your voice notes and try to use the complete set to train a model.
- Unless you're a public figure, or a reclusive billionaire, it's very unlikely that you'll be the target of this kind of cyber attack.
Nonetheless, we do not take any potential threat lightly and we will continue to take steps to monitor and improve content security for all mote users.
What this means for how you use mote:
Before leaving a voice note, ask yourself: would I be embarrassed if this were to be shared beyond the intended audience?
Think about how to use personalization in your messages: using a colleague or student's first name can be a great way to connect. But we'd recommend avoiding using their family name as well.