advertisement
advertisement

Before you agree to any app’s Terms of Service, check this site

Nobody reads terms of service agreements. Guard puts a computer to work reading them for us.

Before you agree to any app’s Terms of Service, check this site
[Illustration: FC]

I don’t need to tell you that we live in a privacy nightmare, in which it seems there’s no social network or app you can truly trust to protect your personal identity and everything you do online. But the worst part is that these companies have us at a disadvantage from the get-go, because they present us with a mountain of legalese—aka the privacy policy of using the service—that we blindly, futilely agree to, all in the name of watching the latest viral lip-syncing video.

advertisement
advertisement

It’s not our fault we do this. No one has the time to outwit a team of legal experts billing who knows what an hour, and few of us have the luxury of just unplugging from our friends, family, and pop culture to remain anonymous. But perhaps there is a way we can fight back against greedy digital companies, with the help of an unlimited resource of our own making: AI.

Guard is a new site built entirely by Javi Rameerez, a one-man-band developer and designer out of Madrid. It presents the apps you know, like Tinder, Netflix, and Instagram, with simple grades rating their privacy. Tap into any of the grades, and you’ll see the criteria: the worst bit of their privacy policy is pulled up for you, in big block letters backed by a cool blue gradient. This is anything but small print at the bottom of a contract; these words are designed to be looked at.

[Screenshot: Useguard]

In the case of Twitter, the concerning excerpt is, “your information may be sold or transferred as part of that transaction” (in reference to selling, merging, or reorganizing the company). In other words, if Twitter is desperate, your personal data is up for a fire sale! In the case of Instagram, the site highlights the passage, “Instagram cannot ensure the security of any information you transmit to Instagram or guarantee that information on the Service may not be accessed, disclosed, altered, or destroyed. Please do your part to help us.” Translation: What’s Facebook, a technology company or something? It’s on you to make sure you say nothing on the service that can be one day be used against you!

Now, maybe a lawyer would tell us that we have little to nothing to worry about when agreeing to these statements. But that’s entirely not the point of Guard. Rather, Guard is built to highlight the things that make everyday people feel the most squeamish, and that’s a reasonable metric to focus on, given that privacy is an inherently personal matter.

“The idea was to evaluate companies’ privacy practices by how they are perceived by the average internet user, not by lawyers or domain experts. Because, you know, there’s nothing illegal about most privacy practices. Lawyers work really hard to protect companies with their legalese. So if you asked a lawyer about any given privacy practice (like, for example, Facebook selling your data to third parties), all they would say is that it’s legal because both parties have agreed to it. And that’s not the point,” writes Rameerez over email. “The challenge was to analyze privacy practices as they are perceived by users. Users don’t like their data being sold. Users don’t like being observed and tracked 24/7. Users don’t like companies knowing private information they’re not even aware of having shared with them.”

[Screenshot: Useguard]

Ideally, if a service has lousy terms of service, you just avoid it altogether. But if nothing else, more transparency in privacy policies offers users, and even federal prosecutors, some means to push back on companies demanding too much information.

advertisement

To actually learn what people are most concerned about, Rameerez built an AI into the site that you can actually help train, just by taking a simple test. The system presents two statements, and it asks which is a better privacy policy. Ideally, with enough answers, the machine learns some ethical rules of thumb and is able to spot the worst offenses in documents of any length. Ideally.

[Screenshot: Useguard]

Today, Rameerez admits on the site that Guard’s results are still “a bit inaccurate” and require more AI training. The holes in the system’s logic are easy to spot. Mozilla—a generally rights-aware nonprofit that offers the web browser Firefox—gets the same “C” rating as Waze, the Alphabet-owned company that tracks your vehicle with so much fidelity that it literally knows the McDonald’s billboards you’re passing along the way. Guard also fails to list any stories under the “scandals” section of Waze, and, well, it probably should.

[Screenshot: Useguard]

But there are good ideas here: namely, that an intellectual democracy can create a machine to battle the technological oligarchy and package it in a design that’s clear enough for any luddite to parse. This is why Rameerez hopes to get privacy experts involved in the project and to secure the sort of funding to take the idea bigger than his own side project.

“I’d love to turn this into a startup that empowers the average internet user with tools to protect their digital privacy,” says Rameerz. “I’d love to turn Guard into something that not only provides knowledge about what’s going on but that also provides the tools to fight back.”

advertisement
advertisement

About the author

Mark Wilson is a senior writer at Fast Company who has written about design, technology, and culture for almost 15 years. His work has appeared at Gizmodo, Kotaku, PopMech, PopSci, Esquire, American Photo and Lucky Peach

More