While showing up at a protest can demonstrate your opinion to the world, you may not want your face — or the faces of other protesters — to be included, especially when there is the possibility that authorities will collect and use that information (as they did for tracking movements during COVID-19 social distancing). As a result, many consider it vital to obscure the faces of people in any photos you may post on social media and other online sources.

What follows are some strategies for removing facial features from your photos. Of course, you can open up your images on a desktop or laptop using Photoshop or Preview to blur or scrub, but we’re going to assume you aren’t carrying around a laptop with you. So with mobile in mind, you still have some solid options.

What needs to be done

When removing faces, you want to use a method that can’t be reversed. It is possible to de-blur a photo, especially using neural networks. It’s not possible to completely reverse the blurring, since it is lossy (in other words, some data will be permanently lost), but a lot can be “restored.” So why take the risk? Painting over faces, or using mosaic blur techniques, will prevent any possibility of reversing the effect.

You also want to remove any and all metadata from your images. They can carry GPS location, timestamps, and details about the type of phone used — basically, lots of things that can be used to pinpoint where you were and when.

Edit out faces

While there are a plethora of apps that will help blur or cover faces and remove metadata for both iOS and Android devices (some of which I mention below), there are ways you can do both without using a third-party app.

First, you can use your built-in photo editor to individually block out faces. On iOS, open Photos, tap on your photo and select the Edit option (at the bottom of the screen). Tap on the Markup icon in the top right of the screen. With that, create solid circles or squares to block out faces. If you have an iPhone that’s series 15 or later with iOS 18.1 (or later), you can also use the Clean Up feature to erase a face: in the photo you want to change, tap Edit and then Clean Up (an eraser icon). Circle the faces you want to obscure, and tap Done.

Android also has a native markup tool — in the Photos app, select the photo, tap on the Edit tool (second from the left on the bottom) and choose Markup (second from the right on the bottom). You can then use the center-bottom Pen tool to scribble over anything you want to cover. You can also (if it’s available on your phone) use the Magic Eraser to erase or camouflage a face: tap on the Edit tool, select Magic Eraser, and circle the face you want to obscure.

Remove metadata

You then want to get rid of the metadata. When you take a photo on your device, meta is going to be attached automatically. The easiest way to avoid this is to take screenshots of your photos so that meta and geotagging won’t carry over. Also make sure to view your photo fullscreen, and ensure you don’t have any notifications or other identifying features in the screenshot.

The same can be done for video. On an iPhone, instead of just using the Camera app, start a screen record while you’re making your video, and use that recording instead. On most Android phones, you can use the Screen record app found in your Quick Settings.

Third-party apps that hide faces & remove meta

Recently, there have been a plethora of apps that will help hide faces and remove metadata for both iOS and Android devices. You may find it easier to use one of these.

For example, encrypted messaging app Signal has a face-blurring tool that is incorporated into its Android and iOS versions.

There are also grassroots efforts like Image Scrubber, which you can use in a browser on your device to upload images to blur and scrub, and then to save the anonymous version back to your device. This is great because it works on all devices, mobile and desktop. 

If you already use apps to edit and enhance photos, you might be able to use those to blur as well. Apps like Glitche (iOS) and Glitch Lab (Android) let you pixelate over selected areas, and Trigraphy (iOS) lets you create mosaic effects.  If you want to take photos now and remove the meta later, you can use the apps mentioned above or photo apps like Halide (iOS) and Snapseed (Android).

In the end, the method you choose will depend on how much work you want to do during or after the fact. For me, I would take photos, edit them in the phone Photo app, take a screenshot — and then delete the originals. Because if your device is unlocked and you have the originals sitting there, then you may have done all that work for nothing.

Update February 26th, 2025: This article was originally published on June 5th, 2020, and has been updated to account for changes in software.

Roku has made my TV unusable.

On June 6th, my TCL TV’s Roku OS was updated to version 13.0.0. Ever since, on everything I watch, there is motion smoothing — a TV and film purists’ deepest nightmare. There is no way to turn it off.

In the updated OS, Roku added a feature called Roku Smart Picture, which, according to its release notes, “automatically improves picture quality dynamically as users stream.” While Roku doesn’t explicitly mention motion smoothing, or what Roku calls “action smoothing,” the update has made it so that I and many others with Roku TVs see motion smoothing, regardless of whether the picture setting is Roku Smart Picture or not. My TV didn’t even support motion smoothing before this. Now, I can’t make it go away.

If you’re someone who doesn’t notice motion smoothing or doesn’t particularly care about it, imagine if, suddenly, your ebook updated so all the fonts were three times as large. Or if your phone decided all video and audio would be played at 2x. Some folks might prefer that, but it should be a choice. Forcing a device to change how a user experiences content that is different than what is expected, with no means to revert or disable the change, is bad. That should be obvious.

Not long after the update rolled out, other Roku TV owners (mainly TCL, but Hisense, too) began posting about the issue in Roku’s community forum and on Reddit. Since I work at The Verge, I told our team about my issue. We reached out to Roku for comment and got no response. We wrote about the problem. Commenters on that post agreed: it sucks. Still, there was radio silence from Roku.

Shipping software can be challenging, especially if you’re shipping updates to a multitude of different devices. QA should catch issues, but sometimes they don’t, and a bug is shipped. It happens! So, in moments like these, it’s important to have strong customer support and to be responsive and communicative. What you probably should avoid is weeks of customers flagging the same issue with no meaningful feedback or updates. Possibly even more important, your support infrastructure shouldn’t be difficult to navigate or have their own bugs that hinder their use. In this case, there’s both.

Unlike my TV’s new picture setting, Roku’s customer support experience has been anything but smooth. I ran into technical issues sending a community moderator my information. The DM button was missing and then magically appeared the next day. Then, I couldn’t send a message, getting an error that my message contained invalid HTML (it did not). I was even told I had “reached the limit for number of private messages that you can send” even though I was never able to send a message to begin with. When I did finally get through, the response I got offered no solution — and barely made any sense.

This whole experience strikes me as something truly wild. If you’re in the business of making a product that plays movies and shows, you should be aware of how divisive a feature like motion smoothing is, as well as how filmmakers feel about it. If your slogan is “happy streaming,” making streaming hell is a bad look. 

With so many competing companies and products in the streaming world, brand loyalty is something every company should covet. Again, bugs happen! We all experience irritating technical issues with apps, sites, and devices. But usually, a brand wants to address those quickly to keep their customers feeling like they are a priority in fear of losing them to any of the many other similar services out there. 

It’s possible that, right now, a team is working on a fix. But I’m not holding out hope: it turns out that this isn’t the first time there’s been an issue with motion smoothing. In 2020, on the same Roku community forum, almost the exact same issue happened to a different group of TCL TVs. The very last comment, posted by a user named DigitalFirefly in 2022, is short and bleak: “I’m still having this issue, I wish they would fix this already.” Me, too, DigitalFirefly. Me, too.

Roku didn’t respond to requests for comment at the time of publish, but we will update this article with any responses we receive.

So, I’m not holding out hope that this will be resolved soon. Because it wasn’t in 2020, and it hasn’t in the last three weeks. Instead, I’m in the market for a new TV, maybe a dumb one that I will never have to connect to the internet. So if you have any recommendations, please share them in the comments.

If broadband access was a problem before 2020, the pandemic turned it into a crisis. As everyday businesses moved online, city council meetings or court proceedings became near-inaccessible to anyone whose connection couldn’t support a Zoom call. Some school districts started providing Wi-Fi hotspots to students without a reliable home connection. In other districts, kids set up in McDonald’s parking lots just to get a reliable enough signal to do their homework. After years of slowly widening, the broadband gap became impossible to ignore.

So as we kick off our Infrastructure Week series, we wanted to show the scope of the problem ourselves. This map shows where the broadband problem is worst — the areas where the difficulty of reliably connecting to the internet has gotten bad enough to become a drag on everyday life. Specifically, the colored-in areas show US counties where less than 15 percent of households are using the internet at broadband speed, defined as 25Mbps download speed. (That’s already a pretty low threshold for calling something “high-speed internet,” but since it’s the Federal Communications Commission’s standard, we’ll stick with it.)

Maps like this are important because, for much of the past decade, the scale of the problem has been maddeningly difficult to pin down. Most large-scale assessments of American broadband access rely on FCC data, a notoriously inaccurate survey drawn from ISPs’ own descriptions of the areas they serve. Even as the commission tries to close the broadband gap, its maps have been misleading policymakers about how wide the gap really is.

Instead of the FCC’s data, we drew on an anonymized dataset collected by Microsoft through its cloud services network, published in increments by the company over the past 18 months. If the FCC monitors the connections that providers say they’re offering, this measures what they’re actually getting. You can roll over specific counties to see the exact percentage of households connected at broadband speed, and the data is publicly available on GitHub if you want to check our work or drill down further. It’s not a perfect dataset, since device speeds can also be affected by bad routers or slow processors that have nothing to do with the underlying connection. But for better or worse, these are the speeds users actually experience, which puts the dataset worlds ahead of what you’d get from the FCC.

The disparity between FCC reports and the Microsoft data can be shocking. In Lincoln County, Washington, an area west of Spokane with a population just a hair over 10,000, the FCC lists 100 percent broadband availability. But according to Microsoft’s data, only 5 percent of households are actually connecting at broadband speeds.

Other areas stand out for the sheer scale of the problem. Nine counties in Nevada fall under the 10 percent threshold, covering more than 100,000 people and the bulk of the area of the state. Most of Alaska is a similar dead zone — understandably, given how rugged the state’s interior is — but similar gaps pop up in southwest New Mexico or central Texas.

Because it’s measuring usage, this data doesn’t distinguish between people who can’t buy a fast connection and people who simply can’t afford one, and in other places, you can see the connectivity problem as one more consequence of accumulated neglect. In Arizona, Apache County stands out as a long thin stripe in the northeast corner of the state, showing just 5 percent broadband usage. More than 70,000 people live there, most of them members of the Navajo, Apache, or Zuni tribes. According to the census, more than 23,000 of them are living in poverty, by far the highest poverty rate in the state. Across the border, San Juan County, New Mexico, shows 29 percent broadband usage, so the problem isn’t that the county is too remote or that the terrain is too difficult to manage. Apache County is simply poor, and the slow progress of the broadband buildout seems like a promise it will stay that way.

 

This story is part of Infrastructure Week

With the right eyes, you can even see the broadband gap as a dividing line for the US at large. Counties on the wrong side of the line are poorer and more remote, losing population even as the country grows. This is why there’s no broadband, of course: from a business perspective, building out fiber in Apache County is a losing bet. But the lack of fiber also stifles economic activity and makes young people more likely to leave, creating a cycle of disinvestment and decay that has swallowed large portions of our country.

In theory, this is a problem the federal government is getting ready to fix. President Biden has proposed $100 billion in broadband funding as part of the American Jobs Plan, more than twice what the FCC estimated would be necessary to bring broadband to 98 percent of households. But it will be a long walk from appropriating that money to actually laying fiber in places like Apache County. That road starts with taking a long look at the shaded parts of this map and thinking about what it will take to truly get them online.

Update 5:11PM ET: Updated to clarify the limitations of device-based speed measurements and its impact on the broader dataset.