Month: November 2017

Posted on

YouTube’s Creepy Kid Problem Was Worse Than We Thought

Image: YouTube / Pexels / Gizmodo

YouTube says that it’s removed ads from some 2 million videos and over 50,000 channels that featured disturbing content aimed at kids. Some of that content actually exploited children in videos. And while we’ve long known that YouTube struggles to keep bad stuff off of its platform, the fact that tens of thousands of channels involved doing bad things to children feels chilling.

The public outcry over YouTube’s creepy kid videos started a few weeks ago. Several reports highlighted seemingly kid-friendly videos that depicted scenes of Disney characters in bikinis flirting with other popular children’s characters and other generally inappropriate themes. The issue was compounded by disturbing videos of cartoon characters dealing with themes like torture and suicide popping up in the YouTube Kids app. There was also a rash of videos that showed kids being tied up, apparently hurt, or otherwise engaged in exploitative situations.

YouTube quickly addressed the issue by announcing plans to age-restrict and demonetize these kinds of videos. The company went beyond that and told Vice News that it “terminated more than 270 accounts and removed over 150,000 videos” as well as “turned off comments on over 625,000 videos targeted by child predators.” Additionally, YouTube says it “removed ads from nearly 2 million videos and over 50,000 channels masquerading as family-friendly content.”

News of YouTube’s action against millions of disturbing videos came on the heels of a separate but related controversy. Over the weekend, a number of people reported that typing “how to have” into the YouTube search bar prompted autofill suggestions that include phrases like “how to have s*x with kids” and “how to have s*x in school.” Those searches led to videos with titles like “Inappropriate Games You Played as a Kid” featuring a provocative thumbnail of young people kissing and “School Is Hard and So Is Your Math Teacher” with an image of a crying girl being touched by an older man. Those videos have 23 million and 117 million views, respectively, so it’s not hard to imagine why they showed up at the top of the search results.

YouTube says it has removed these vile suggested searches, which is good. But the persistence of these kinds of problems raises larger questions about YouTube’s capacity for moderating creepy content. I’m not talking about “spookin in the wrong neighborhood” or any other fun but weird but also slightly dark videos. I’m talking about the stuff that targets kids and appeals to bad people, like pedophiles.

The problem with all of these videos is how borderline they appear to be—even to humans. And yet, YouTube primarily depends on algorithms and filters to keep bad content off its platform. Videos and channels are removed by human moderators, but only after they’re flagged by users. In the meantime, you have disturbing autofill suggestions like “how to have s*x in school” showing up for everyone, as well as the countless questionable videos to which these searches lead. Removing all of these suggestions and videos seems like an impossible task, especially since over 400 hours of content are uploaded to YouTube every minute.

The thing is, algorithms are inherently imperfect. Computers have a hard time identifying the uncanny valley that separates an innocuous video from one that’s entirely inappropriate. Sure, videos that violate YouTube’s terms of service—stuff that’s copyrighted, gruesome, illegal, full of nudity, exploits children, and so on—can get flagged and removed. YouTube also uses algorithmic filters now in order to catch some of these videos before they’re published. The system isn’t perfect, but YouTube seems committed to it.

It’s inevitably hard to point fingers. Is it a tech company’s fault that humans are awful and abusive and exploitative? Of course not. YouTube does shoulder a tremendous burden when it comes to deciding how to let the right videos in and keep the bad ones out. Few are surprised when the platform fails to catch every creepy video. But the creepy videos are still a problem. Whether that says more about YouTube’s limitations or our own perversions remains to be seen.

from Gizmodo http://ift.tt/2AdazcE
via IFTTT

Posted on

Photoshop uses AI to make selecting people less of a hassle

Masking a human or other subject out of a scene is a pretty common trick nowadays, but it’s is still arguably one of the hardest and lowest-tech parts of Photoshop. Adobe’s about to make that a lot easier, thanks to an upcoming AI-powered feature called Select Subject. Using it is pretty much idiot-proof: From the main or "Select and Mask" workspaces, you just need to click anywhere on the image, and it’ll automatically select the subject or subjects in the image. From there, you’re free to change the background or tweak the subject separately.

The tech is powered by Adobe’s AI platform, Sensei. "Complicated details around the subject aren’t an issue, because this feature is using machine learning to recognize the objects," Adobe Photoshop Product Manager Meredith Payne Stotzner says in the YouTube video (below). During the demo, she uses it select a single person on the street, a group of volleyball players, a couple on the beach and a red panda.

In some cases, details like hair and fur aren’t properly selected, but using the tool would certainly give you a big head start. It pretty much eliminates the tedious hand-drawn or tweaked masking process, letting you focus on the fine details. Since it uses machine learning tech, it should also get better over time.

There are already plenty of Photoshop plugins like Akvis SmartMask and Fluid Mask that can do something similar to Select Subject. However, it’ll be nice to have such a feature as part of Photoshop, rather than paying extra for a plugin. And the new feature is more than just a technology "sneak" — it’s an actual feature coming in a future Photoshop build. Adobe has yet to say exactly when it’ll arrive, however.

Source: Adobe (YouTube)

from Engadget http://ift.tt/2ncHIAj
via IFTTT

Posted on

Airbnb now lets you split the cost of rentals

Airbnb is taking a page out of Uber’s playbook.

The short-term rental startup has announced it will now let users split payments on the platform.

Airbnb users can split the cost of a listing with up to 16 people. Previously, the trip organizer would have to pay for the entire cost of the stay upfront.

Payments are automatically split evenly as a default. However, travelers can choose to pay for more than one person.

Related: Airbnb-branded apartment building to open near Walt Disney World

Last year, Airbnb CEO Brian Chesky asked users on Twitter what feature they’d most like to see in 2017. One of the top requests was the ability to split payments.

The tool will launch worldwide on Tuesday, with the exception of India and China. The current payment rules and regulations in those countries would make it difficult to roll out the feature, according to an Airbnb spokesman.

The option is a part of Airbnb’s greater effort to attract more users. During a test of the feature, the startup said 30% of reservations booked led to one or more new users joining Airbnb. It recently tested the feature among 80,000 groups across 175 countries.

But it’s hardly a new concept. Ride-hailing startups Uber and Lyft both offer a split fare option. Uber said its split fare tool is used during “hundreds of thousands” of rides every month, making it one of its more popular features.

Related: You can now book a restaurant reservation on Airbnb

The tool comes a few months after the company rolled out restaurant reservations within its app. Meanwhile, Airbnb announced earlier this month it is acquiring startup Accomable to help connect disabled travelers with more accessible listings.

The platform is also trying to expand beyond its business model. Earlier this year, it announced plans to build a 324-unit apartment building in Kissimmee, Florida in early 2018.

CNNMoney (New York) First published November 28, 2017: 9:02 AM ET

from Business and financial news – CNNMoney.com http://ift.tt/2jtJ0Sq
via IFTTT

Posted on

Google Waymo’s autonomous cars have driven 4 million miles on public roads

Lest anyone think that Waymo hasn’t been preparing to launch its own autonomous ride-sharing service at some point, the Google spinoff just announced that its self-driving cars have driven a collective 4 million miles on public roads. But it’s not just the milestone the company is celebrating, it’s the pace: While it took the company 18 months to reach 1 million, then 14 to reach 2 million, then 8 months to reach 3 million and finally six months to reach the 4 million mile marker.

While these have been on public roads, they’ve been restricted to cities in California’s Bay Area and around Phoenix, Arizona, with some testing in Austin, Texas and Kirkland, Washington. Waymo had been refining its autonomous tech in partnership with Lyft, though it’s unclear how much influence that has had.

In addition to real driving, the company has put its systems through 2.5 billion simulated miles in the last year, Waymo’s blog post boasted. With all that experience, the company says its been able to teach its vehicles enough to pull off “full autonomy,” and hinted that the public would soon “get to use Waymo’s driverless service to go to work, to school, to the grocery store and more.”

Waymo Blog

Written by David Lamb for Engadget.

Related Video:

from Autoblog http://ift.tt/2AEXVEm
via IFTTT

Posted on

Watch Out: If Someone Blocks You on Twitter, You Lose Your DM History With Them

The Twitter block is a powerful tool. Whereas muting is a subtle cleansing of your timeline—the person you mute never knows, though you have to wonder, how many people might have muted you??—blocking is Twitter’s carpet bomb.

When you block someone:

  • They cannot see your tweets. You cannot see theirs.
  • You unfollow each other.
  • If they try to see your profile, they will see that you’ve blocked them.
  • They cannot DM you.

For the full repercussions, here’s Twitter’s guide. Right now, I want to focus on the DMs issue. Because this was recently brought to my attention:

Blocks happen for all sorts of reasons. Maybe someone was being abusive. Maybe muting wasn’t strong enough, and you wanted to fully protect your timeline from someone’s presence. Or maybe you got in a Twitter fight—in public or in DMs—and one of you gets fed up and pulls the trigger. If that happens, it’s useful to know that any DM history from before the blocking will be erased. Be safe: take screenshots, keep your receipts.

from Lifehacker http://ift.tt/2n4bPKf
via IFTTT

Posted on

Imgur Was Hacked in 2014 Exposing 1.7 million User Accounts

Posted by

Shane McGlaun |

Mon, Nov 27, 2017 – 9:00 AM

Back in 2014 Imgur was hacked and one of the scariest parts of this hack was that it had no idea the breach had occurred until this month. The hack resulted in the usernames and passwords of 1.7 million accounts being stolen. Another troubling thing about the hacks is Imgur has no idea how exactly the hackers gained access.

Imgur wrote, “We are still investigating how the account information was compromised. We have always encrypted your password in our database, but it may have been cracked with brute force due to an older hashing algorithm (SHA-256) that was used at the time. We updated our algorithm to the new bcrypt algorithm last year.”

Imgur says that it asks for no personally identifiable information to open an account. The users affected by the branch have been contacted on the email addresses that they used when registering. Those users are being required to change their passwords.

The security researcher who discovered the breach is Troy Hunt and he was given a list of the stolen account details by an unnamed source reports PCMag. Hunt told ZDNet, “I disclosed this incident to Imgur late in the day in the midst of the US Thanksgiving holidays. That they could pick this up immediately, protect impacted accounts, notify individuals and prepare public statements in less than 24 hours is absolutely exemplary.”

from Legit Reviews Hardware Articles http://ift.tt/2zGCtPj
via IFTTT

Posted on

Catching lightning in a volcanic bottle

What do you get when you marry two of Earth’s most dramatic natural events, lightning and volcanoes? The answer is a "dirty storm," an infernal melange of lightning, magmatic fire and ash that surpasses even the wildest Hollywood disaster movie effects. If the volcano has enough energy, dozens of bolts — which blast upwards, rather than downwards — can spawn in the ash. At at the Volcán Calbuco in Chile, photographer Francisco Negroni captured a stunning example of the phenomenon, winning second prize at this year’s Epson Pano awards.

If you think I’m exaggerating about how dramatic these storms can look, a quick Google image search of "dirty storm" will change your mind. All suggest that some kind of impending apocalypse/horsemen scenario is about to take place.

Many of the best ones were taken by Negroni himself, but even he admits that the image above is special. "The photograph with which I obtained second place in the Pano Awards was taken during the violent eruption of the Volcán Calbuco, located in the south of Chile," he told Engadget via email. "The technique was simple: long exposure, tripod and a 80-200mm lens. Approximately 10 minutes to achieve that incredible image that, without a doubt, is my best photograph of an eruption and I think the best taken in the world."

How does this happen? In a regular thunderstorm cloud, lightning is created when rising air makes ice crystals and water droplets bump together, forming static electricity. Once the charges build up enough to surpass the atmosphere’s natural insulating tendency, lightning discharges either from one cloud to another or to the ground.

Volcanic lightning works on the same principle, but via different actions. Recently, researchers studying one of the world’s most active volcanoes, Japan’s Mount Sakurajima (above), learned more about how it forms. Using high-speed cameras, they discovered that charge created by churning magma builds around the rim, electrifying the ash just above it.

That creates an electrostatic potential in the lower portion of the ash that eventually causes lightning to discharge into the cloud or air, often in the opposite direction of regular lightning. As such, lightning formation is usually limited to the bottom part of the ash plume, and depends strongly on how the plume develops.

Chile is located on the Pacific "Rim of Fire," and has 90 active volcanoes, the second most in the world after Indonesia. Negroni has photographed volcanoes like Llaima in the Araucania region, the 2011 eruption of Puyehue-Cordon Caulle, and in 2015, the Villarrica Volcano and Calbuco pictured here. Volcanologists and seismologists have become extremely adept at detecting eruptions ahead of time, so endangered inhabitants were evacuated well ahead of time.

Though volcano science has improved dramatically, getting near active ones that produce lightning isn’t for the faint of heart. "Photographing volcanoes is very dangerous and I do not recommend it to anyone," Negroni said. "But if you have the opportunity to do it from a very safe place, it will be an incredible show that nature gives us."

Via: Kottke

Source: Francisco Negroni

from Engadget http://ift.tt/2zt3K3C
via IFTTT