By Lily Hay Newman for WIRED.
As end-to-end encrypted messaging apps have exploded in popularity, several well-known services have added encrypted calls as well. Why not, right? If it works for text-based chat, voice seems like a natural extension. If only it were that easy.
Encrypting calls has plenty of value, keeping conversations strictly between the two parties. They can circumvent government wiretaps, or criminal snooping. But a host of technical challenges with facilitating the calls themselves has slowed the spread of voice over internet protocol overall. Bandwidth is expensive. Firewalls and network filters make it harder to route data streams. Even basic call quality issues, like delays and echoes, prove difficult to fix. Adding encryption on top of all of this takes additional resources and specialized developers.
All of which has delayed encrypted calling — but not stopped it. And a new groundswell of enthusiasm is bringing more options than ever.
The challenges of making reliable encrypted calling starts with the underlying premise of internet-based calls. They’re hard. While VoIP calling has become more reliable over the years, it remains technically challenging in itself, especially when people use cellular data instead of more stable ethernet or Wi-Fi connections.
Despite those challenges, Signal, the well-regarded secure communication platform, has offered encrypted calling since 2014. And when WhatsApp followed in 2016, bringing encrypted calls and video chat to more than a billion users, it helped shake off some longstanding inertia. Other secure messaging apps like Wire and Telegram have added encrypted calling over the last year. Signal itself even rolled out call quality improvements in February.
Signal developer Open Whisper Systems open-sources its code, so that companies can borrow from it to build their own encrypted chat and calling features. For example, while WhatsApp’s overall setup is proprietary, it bases the key exchange for its end-to-end encrypted messages and calls on Signal Protocol. Its users have to trust that it is implementing true end to end encryption in the way it claims. In exchange it brings some form of end to end encryption to an enormous user base that would probably otherwise have little exposure to or protection from the feature. And customers who don’t have faith in a large provider like WhatsApp now have other options, given the recent proliferation of both VoIP in general and encryption specifically.
“There’s so much happening right now in this space which is really exciting,” says Nathan Freitas, the founder and director of the Guardian Project, a privacy and security nonprofit that worked on an encrypted calling platform called Open Secure Telephony Network. “In 2012 there was just Skype basically. Google Hangouts didn’t even exist. FaceTime existed kind of. So we’re really happy when there’s so much public innovation that includes privacy and security.”
Though not nearly as much as there could be, if everyone could get on the same page.
As with messaging, end-to-end encrypted calls require that both ends of the conversation use the same system. In other words, using Signal to call a landline won’t cut it; you need to connect with another Signal user. Given this reality, many developers naturally gravitate to implementing encryption in closed systems; it’s easier both to manage and monetize.
For users, though, this approach has downsides. Unless the developer makes the product fully open source, or allows for extensive independent auditing, there’s no guarantee that the encryption implementation works as advertised. The lock-in factor also limits who you can safely communicate with, which slows adoption.
Imagine, instead, an open communication standard that includes end-to-end encryption. It would allow secure communication with more people between different products and interfaces, because the protocols facilitating the end to end encryption would be the same.
The Guardian Project’s OSTN experiment attempted to create exactly that sort of comprehensive, open communication suite. It focuses on using existing open, interoperable communication standards, employing classic protocols like ZRTP, which was developed in the mid 2000s by PGP creator Phil Zimmerman, and SRTP, which was developed in the early 2000s at Cisco. It also coordinates and controls its voice calls using the Session Initiation Protocol, developed by the telecom industry in the mid 1990s.
That retro backbone didn’t come by choice; there simply aren’t a lot of more modern open protocol options available. Most big VoIP plus encryption advances have come from private companies like Skype (now owned by Microsoft), Google, and Apple, who offer varying degrees of encryption protection for calls and tend to value locked-in users over interoperability. That left OSTN with old tools.
“While they’re very powerful, these are things that are 10, 20, 30 years old in terms of the architecture and the thinking,” Freitas says. “They’re definitely showing their age.”
And while a few smaller services, like PrivateWave and Jitsi, have adopted OSTN, the decision by larger companies to go it alone has limited its open-protocol dreams. That’s especially a shame for people who need absolute guarantees of security.
Rolling Your Own
With proprietary apps, it can be hard for a user to tell if end-to-end encryption is enabled on both ends. Or, in the case of apps whose encryption protocols have not been fully vetted, whether it works as advertised to begin with.
“For mainstream services, crypto is a nice add-on to give users the idea that they can feel more secure, but that’s completely different than when your [customers] are people who are under threat,” says Bjoern Rupp, the CEO of the boutique German secure communication firm CryptoPhone. “If you have to fear for your life, not all secure communication systems are designed for that.”
Encryption die-hards can host their own system using open standards like OSTN, similar to how you might host your own email server. Though it takes some technical knowhow, it’s an option that gives users real control and that isn’t possible with closed systems. Another option is to use a security first service like CryptoPhone that offers an integrated, one-stop solution.
CryptoPhones can only call other CryptoPhones, but the company made that choice so it could control the security and experience of both hardware and software. To reconcile this closed system with transparency, the company is open source and invites independent review. It also has over a decade of experience. “CryptoPhone has been making high-end commercial products for secure voice calling for a long time,” the Guardian Project’s Freitas says. “They had these crypto flip phones, which were awesome.”
None of which leaves the average consumer with widespread encrypted calling that works across multiple services. There may be some help on the way, though, in the form of a new, open, decentralized communication standard called Matrix that includes end to end encryption for chat, VoIP calling, and more. Matrix could be a clean, easy to implement standard underlying other software. For instance, if Slack and Google Hangouts both used the Matrix standard, you would be able to Slack someone from Hangouts and vice versa, similar to how you can send emails to anyone using their email address, regardless of what provider they use.
“The net owes its existence to open interoperability,” says Matthew Hodgson, technical lead of Matrix. “Then people build silos to capture value, which is fair enough, but you get to a saturation point where the silos start really stifling innovation and progress through monopolism.”
The catch, of course, is getting buy-in from companies that have little incentive, or getting new services built on a standard like Matrix to take off. Walled gardens tend to produce more profit than open ones.
Still, having these new options is an important first step. And combined with the broader proliferation of encrypted voice-calling apps, change finally seems to be coming from a lot of directions at once. “I think there’s a longer-term project going on called the internet,” Freitas says. “Some of us still believe in it.”
More from WIRED: