Category: Tech news

Self-driving startup Waymo, a Google spin -off owned by parent company Alphabet, has started to test pricing models for rides in its autonomous vehicles in Phoenix, the latest indication that the company is preparing to launch a commercial robotaxi service.

Waymo has not launched a wide-scale commercial robotaxi service in Phoenix — or anywhere — just yet. But it is getting closer.

Waymo’s early rider program, designed to give a vetted group of real people the ability to use an app to hail a self-driving vehicle, has been expanded, Alphabet CFO Ruth Porat explained Thursday during the company’s quarterly earnings call. Waymo started testing pricing models within its app during the third quarter, Porat said.

The early rider program had 400 participants the last time Waymo shared figures on the program. A Waymo spokesperson declined to elaborate on how much it had grown.

“As part of our early rider program, we have recently begun testing pricing models within our app,” a Waymo spokesperson said in an emailed statement. “Pricing is currently experimental and intended solely to solicit feedback from early riders and does not reflect the various pricing models under consideration for a public service.”

Waymo has been inching toward a commercial service in Phoenix since it began testing in the suburbs like Chandler in 2016. It started in earnest when Waymo launched the early rider program in April 2017. Later that year, Waymo removed employees and passengers from its test fleet, sending empty self-driving minivans onto the streets of greater Phoenix.

By May of this year, Waymo began allowing some early riders to hail a self-driving minivan without a human test driver behind the wheel. More recently, the company launched a public transit program in Phoenix focused on delivering people to bus stops and train and light-rail stations.

Testing continues in other cities as well, including Mountain View, California and Austin. The company announced earlier this month that its autonomous vehicles have driven 10 million miles on public roads in the United States.

At least two Twitter accounts linked to the man suspected of sending explosive devices to more than a dozen prominent Democrats were suspended on Friday afternoon.

Cesar Sayoc Jr., 56, was apprehended by federal law enforcement officers in Florida on Friday morning. “Though we’re still analyzing the devices in our laboratory, these are not hoax devices,” FBI Director Christopher Wray said during a press briefing.

Facebook moved fairly quickly to suspend Sayoc’s account on the platform, though two Twitter accounts that appeared to belong to Sayoc remained online and accessible until around 2:30 p.m. Pacific. Both accounts featured numerous tweets, many of which contained far-right political conspiracy theories, graphic images and specific threats.

TechCrunch was able to review the accounts extensively before they were removed. Both known accounts, @hardrockintlet and @hardrock2016, contained many tweets that appeared to threaten violence against perceived political enemies, including Keith Ellison and Joe Biden, an intended recipient of an explosive device.

In one case, those threats had been previously reported to Twitter. Democratic commentator Rochelle Ritchie tweeted that she reported a tweet from @hardrock2016 following her appearance on Fox News. According to a screenshot, Twitter received the report and on October 11 responded that it found “no violation of the Twitter rules against abusive behavior.”

The tweet stated “We will see u 4 sure. Hug your loved ones real close every time you leave home” accompanied by a photo of Ritchie, a screenshot of a news story about a body found in the Everglades and the tarot card representing death.

Update: Twitter issued an apology for not dealing with Ritchie’s initial report.

Between the two accounts linked to Sayoc, many of the threats were depicted with graphic images in sequence. In one tweet on September 18 to former Vice President Joe Biden, the account tweeted images of an air boat, a symbol depicting an hourglass with a scythe and graphic images of a decapitated goat.

Threatening messages that emerge out of a sequence of images would likely be more difficult for machine learning moderation tools to parse, though any human content moderator would have no trouble extracting their meaning. In most cases the threatening images were paired with a verbal threat. At least one archive of a Twitter account linked to Sayoc remains online.

In a statement to TechCrunch, Twitter stated only that “This is an ongoing law enforcement investigation. We do not have a comment.” The company indicated that the accounts were suspended for violating Twitter’s rules, though did not specify which.

To keep up with the rising demand for short-term rentals in U.S. cities and compete with the home-sharing giant Airbnb, travel booking site Expedia has picked up a pair of venture-backed hospitality startups, Pillow and ApartmentJet.

Employees of both companies will join Expedia . The company declined to disclose the financial terms of the deals.

“Acquiring Pillow and ApartmentJet will help unlock urban growth opportunities that, over time, will contribute to HomeAway’s ability to add an even broader selection of accommodations to its marketplace and marketplaces across Expedia Group brands, ensuring travelers always find the perfect place to stay,” the company explained in a statement.

Expedia paid $3.9 billion for HomeAway and its portfolio of travel brands in 2015. The deal was its first major move in the alternative accommodations space, as well as the beginning of a series of efforts to outdo VC darling Airbnb. Its latest targets provide software tools for property managers to easily manage short-term rentals on Airbnb competitors like HomeAway and VRBO.

Located in San Francisco, Pillow helps residents list their apartments as short-term rentals without violating their leases. It’s raised a total of $16.5 million in VC backing since 2013, including a $13.5 million round last year led by Mayfield, with participation from Sterling.VC, Peak Capital Partners, Expansion VC, Chris Anderson, Gary Vaynerchuk, Dennis Phelps and Veritas Investments.

ApartmentJet helps property owners earn money off vacancies. Founded in 2016, the Chicago-headquartered startup had raised a reported $1.2 million in capital from Network Ventures and BlueTree.

Bellevue-based Expedia Group owns several travel brands, including HomeAway, VRBO, Travelocity, trivago, Orbitz and Hotels.com. The company is both an active investor in and acquirer of startups.

Expedia’s shares rose 9.4 percent Thursday after its third-quarter earnings beat analyst expectations. The company posted $3.28 billion in revenue, a notable increase from last year’s $2.97 billion.

The question of whether a machine can create art, or anything at all, is at the heart of many a philosophical debate and has been for decades. But whether it’s worth something on the market? That point has been settled definitively today as a portrait-like image issuing from an AI sold for nearly half a million dollars at auction.

“Edmond de Belamy,” whom you see above, such as he is, is one of several members of a fictitious family created by a “generative adversarial network,” in turn created by French AI engineers and artists Obvious.

GANs comprise two parts, for which terminology differs but Obvious calls the “generator” and the “discriminator.” Both visual recognition models are given a set of data to ingest, in this case 15,000 portraits from the last 600 years or so. Based on this data, the generator attempts to create new portraits, and the discriminator tries to identify those portraits as either authentic or artificial. The less sure the discriminator is that an image is artificial, the closer it tends to be to the authentic portraits.

The Belamy family is the result of this process playing out many times, producing the strange, distorted faces that have a dreamlike, and also nightmarish, quality to them.

They’re also unmistakably computer-generated. The patriarch and Count of the family, for instance, though the colors and gross figure are interesting and in broad strokes painterly, the pattern of stippling (or whatever you want to tall it) is a telltale mark of a computer attempting to create consistent texture. His wife, the Countess, has a psychedelic oil-slick quality to her hair and dress that’s quite unnatural, and what appears to be craquelure on closer inspection is revealed to be an intricate warping structure reminiscent of Photoshop effects.

“It is an attribute of the model that there is distortion,” explained Hugo Caselles-Dupré, from Obvious, to Christie’s. “The Discriminator is looking for the features of the image — a face, shoulders — and for now it is more easily fooled than a human eye.”

Obviously it doesn’t quite match the old masters. But as you can see from the variety evinced by the Belamy clan, the system has a remarkable range and one can intuitively grasp the type of painting this is — perhaps each even reminds you of a real one.

Certainly someone thought that Edmond at least was worth having; Obvious estimated that the painting (though surely a print) would fetch perhaps €10,000 on the block. Imagine the group’s surprise when the bidding escalated to $432,500 — obviously $500 too much for one of the bidders. The new owner remains anonymous, though we may learn more later. For all we know it is Obvious itself (unlikely) or some art holdings company speculating that this early AI piece may become an historic one.

As for the signature, a rather tongue-in-cheek solution was lit on by the team: At the bottom right of Edmond’s canvas is part of the algorithm that created him (though far from all the code required to do so).

The work page is a bit more specific: “generative Adversarial Network print, on canvas, 2018, signed with GAN model loss function in ink by the publisher, from a series of eleven unique images, published by Obvious Art, Paris, with original gilded wood frame.”

We’re no closer to getting at the heart of art, deciding whether these generated constructs count as art, and if so, by whom, but it’s interesting nevertheless. Even if these aren’t exactly the kind of thing you’d want to hang on your wall. That’s true of most art anyway.

Avery Haskell says he first knew he wanted to be an astronaut ever since he was a boy growing up in Houston near NASA’s Johnson Space Center.

The 24-year-old Stanford graduate who counts Stephen Hawking and Carl Sagan as his heroes grew up in an entrepreneurial family. In the early days of the internet his mother, an accountant in the oil and gas industry, and father, an information technology technician for a railroad, launched their own startup called “Neighbornet” — an early version of Zillow (which never got off the ground).

Haskell himself bounced around the startup industry, with forays into launching a crowdfunding startup and stints at a few mobile technology companies, before landing on his current venture, Shuttle.

Launched earlier this year out of the Alchemist Accelerator and co-founded with cybersecurity expert and Wickr co-founder, Nico Sell, Shuttle is aiming to be the web and mobile-based booking agent for spaceflight.

“Space is my first love,” said Haskell, who helped found the Stanford Space Initiative at his alma mater. “I’ve always wanted to be an astronaut and help more people become astronauts. I thought it would be cool to get more people to go to space and get more people interested in space travel.”

Haskell met Sell at the Alchemist Accelerator, where she first worked as a mentor to the young entrepreneur. But she quickly became enamored with the idea of working at the edge of a new kind of frontier market. The day that Sell agreed to be the chair of Shuttle was the day Elon Musk’s SpaceX landed two booster rockets back on earth nearly simultaneously.

“I’m following Elon into space,” said Sell. “When I first started working with Avery I had asked ‘Are we really ready for that now?’ And after working with him I’m convinced that we are.”

Purchasing tickets on a flight listed on Shuttle isn’t the same as buying a plane ticket on Kayak, primarily because the price points are higher to the point of near-absurdity if you’re not a member of the super rich.

Offerings will range from trips on Virgin Galactic trips that will cost upwards of $250,000 in the near term to low-end packages that will include a zero-gravity flight aboard a tricked-out Boeing 747 for the low, low price of just under $5,000 per seat.

The company is actually taking orders for its first zero gravity flight, which it expects to launch from San Francisco in March 2019. That’ll give roughly 34 people the opportunity to experience weightlessness for around 8 minutes.

“Our mission is to open space up to everyone,” says Haskell. “We want to get more people to space so that the price goes down and so that more people can see earth from space and become private astronauts.”

Eventually, as more space tourism offerings become available, the company expects to sell additional packages. “There’s a luxury space hotel that’s being built right now,” says Sell. “It’s a million dollars a night and a 12 night minimum and every 90 minutes you see a sunset and a sunrise. Pretty soon there’s going to be a moon walk and a space walk that are available too.”

Shuttle is hoping to be the hub that aggregates all of these offers into a single one-stop shopping and media experience for consumers interested in seeking out existing planets and boldly going where only few men (and women) have gone before. And the company will offer virtual space tours and tickets to launches for the plebes who can’t afford an actual ride.

Initially, expect the ultra-rich or the ultra-subsidized to be the only folks that will be able to take these trips. Sell sees a lot of opportunities in corporate packages for business customers — likening it to a trip to Kelly Slater’s Surf Ranch for an executive retreat.

Sell believes that there will be upwards of 100,000 people in the next 10 years who’ll be willing to plunk down the $50,000 to $250,000 that it will cost to go to space.

Already, the company has $1.66 million in bookings off of 8 customers on four Virgin Galactic flights and four Zero Gravity Charters with commission rates of 5 percent to 10 percent on flights that average $250,000 per ticket.

As for what comes next, Haskell has some speculations. “We will probably be able to build a base on the moon very soon. By 2030 that’s a possibility. Within my lifetime it will be pretty common for people to travel to and from other planets in space,” he said. 

For him, the importance of Shuttle is getting Earth’s human residents to realize the fragility of our existence on the tiny blue ball we all share. Haskell said his favorite quote from Carl Sagan was “We are a way for the cosmos to know itself.” And if that’s true, Haskell believes that the experience of traveling through the cosmos may be a way for humans to come to a better understanding of themselves as well. 

In recent years, many have pushed to level the playing field for women in tech through new initiatives, funds, companies, support networks and more. White women, however, have emerged as the key beneficiaries.

Less has been done to bolster black and Latinx female founders specifically. Enter digitalundivided. The organization, which encourages black and Latinx women to pursue entrepreneurship, has been working to highlight just how extensive the disparity in funding is for black and Latinx female founders.

Today the company published its first-ever report on venture capital funding, or lack thereof, for Latinx female founders via its proprietary research arm, called ProjectDiane, in partnership with JPMorgan Chase.

The key takeaways: Latinx women make up 17.1 percent of the U.S. women’s population but less than 2 percent of women-led startups have Latinx women founders. Of the 107 Latinx women-led startups in digitalundivided’s database, 58 of them have raised more than $1 million in funding. Overall, Latinx women have raised only .4 percent of the more than $400 billion VC dollars invested in startups since 2009. 

“We are proud to be continuing the push toward a world where all women own their work through entrepreneurship, because that’s the path to real power and economic stability for black and Latinx communities,” said Kathryn Finney, digitalundivided’s chief executive officer, in a statement. “digitalundivided understands the impact of data on policy and startup ecosystems which is why we’re committed to using ProjectDiane to further develop data-driven programs for black and Latinx women founders and shape the narrative about women of color in startups.”

The company has released several reports on VC funding for black women founders. Their latest showed a slight increase in the number of black women to raise significant funding rounds. Still, black women have raised just .0006 percent of all VC since 2009.

In 2017, U.S. female-founded companies raised 2.2 percent of all VC, an abysmal and highly cited statistic. That number looks to be increasing ever-so-slightly in 2018, but the industry has a long way to go before capital is equally distributed.

The Pentagon’s $10 billion JEDI cloud contract bidding process has drawn a lot of attention. Earlier this month, Google withdrew, claiming ethical considerations. Amazon’s Jeff Bezos responded in an interview at Wired25 that he thinks that it’s a mistake for big tech companies to turn their back on the U.S. military. Microsoft president Brad Smith agrees.

In a blog post today, he made clear that Microsoft intends to be a bidder in government/military contracts, even if some Microsoft employees have a problem with it. While acknowledging the ethical considerations of today’s most advanced technologies like artificial intelligence, and the ways they could be abused, he explicitly stated that Microsoft will continue to work with the government and the military.

“First, we believe in the strong defense of the United States and we want the people who defend it to have access to the nation’s best technology, including from Microsoft,” Smith wrote in the blog post.

To that end, the company wants to win that JEDI cloud contract, something it has acknowledged from the start, even while criticizing the winner-take-all nature of the deal. In the blog post, Smith cited the JEDI contract as an example of the company’s desire to work closely with the U.S. government.

“Recently Microsoft bid on an important defense project. It’s the DOD’s Joint Enterprise Defense Infrastructure cloud project – or “JEDI” – which will re-engineer the Defense Department’s end-to-end IT infrastructure, from the Pentagon to field-level support of the country’s servicemen and women. The contract has not been awarded but it’s an example of the kind of work we are committed to doing,” he wrote.

He went on, much like Bezos, to wrap his company’s philosophy in patriotic rhetoric, rather than about winning lucrative contracts. “We want the people of this country and especially the people who serve this country to know that we at Microsoft have their backs. They will have access to the best technology that we create,” Smith wrote.

Throughout the piece, Smith continued to walk a fine line between patriotic duty to support the U.S. military, while carefully conceding that there will be different opinions in a large and diverse company population (some of whom aren’t U.S. citizens). Ultimately, he believes that it’s critical that tech companies be included in the conversation when the government uses advanced technologies.

“But we can’t expect these new developments to be addressed wisely if the people in the tech sector who know the most about technology withdraw from the conversation,” Smith wrote.

Like Bezos, he made it clear that the company leadership is going to continue to pursue contracts like JEDI, whether it’s out of a sense of duty or economic practicality or a little of both — whether employees agree or not.

Google Docs just rolled out a time-saving trick that’s sure to be welcomed by heavy users of Docs, or any of Google’s other productivity tools like Sheets, Slides, Sites or Forms. The company this week introduced its  “.new” domain, which can be used to instantly create a new file across any of these services, it says.

For example, instead of going to Google Drive, clicking the “new” button, then the service you want to use, you can just type “doc.new” to get started in a new Google Doc.

Google helpfully registered many variations on this domain, as well, so docs.new and documents.new also work.

And the same format applies across Google’s productivity apps, meaning you can also type in things like sheet.new, sheets.new, spreadsheet.new, site.new, sites.new, website.new, slide.new, slides.new, deck.new, presentation.new, form.new or forms.new.

(Don’t type in the “www” — just the domain.)

If you tend to work with Google Docs on a regular basis, this little hack can end up saving a ton of time throughout your day. You can even bookmark the domains to use as shortcuts, so you can get to the same blank document with just a click.

This is all possible because Google owns the .new domain, which allows it to create whatever subdomains it wants on the site.

After Google tweeted the news on Thursday, users were so thrilled about the trick they started requesting other domains, too. “Do drawings pretty please,” asked one Twitter user. “Please also add email.new,” said another.

Google didn’t respond to those requests, but it wouldn’t be surprising if the domain was used in other ways across its apps in the future.

Grin, the Mexico City-based electric scooter company backed by Y Combinator, is merging with São Paulo-based Ride to further the company’s expansion across Latin America. This comes shortly after Grin raised a ~$45 million Series A round.

Currently, Grin only operates in Mexico City, but it has plans to expand to other cities throughout Latin America. The merger with Ride, which already operates in São Paulo, will enable Grin to do this as early as next week, Grin co-founder Sergio Romo told TechCrunch.

As part of the merger, Ride will operate under the Grin brand in Brazil and the Ride team will be in charge of all of Grin’s operations in Brazil. Ride is currently the only shared electric scooter operator in all of Brazil, but that will soon change when Yellow deploys its scooters. Last month, Yellow raised a $63 million Series A round for its bike- and scooter-share company.

Grin has also partnered with Colombia-based Rappi, an on-demand delivery startup that raised $200 million back in August. This partnership, which will enable Rappi customers to unlock Grin scooters through the Rappi app, will help boost Grin’s expansion across Latin America, Romo said.

While LATAM is a huge market, Grin ultimately envisions operating its pick-up and drop-off scooter model worldwide.

“We definitely want to be global,” Romo said. “I don’t think you can become a ten-billion-dollar company if you don’t go global. I think LATAM might actually be the best market — there’s huge density and a huge market combined with Europe. And who knows, we might pop up in an American city soon if we do a good job. But this is definitely in our heads. This is engineered to be a global play.”

California’s much-anticipated net neutrality rules, which were signed into law last month, are being put on ice until a challenge to the FCC’s own rules at the federal level is resolved. It’s unfortunate, but logical — if the FCC rules are undone or modified, the necessity and legality of California’s will also be affected.

As you likely remember, the FCC repealed 2015’s net neutrality rules at the end of 2017 and implemented a new, much weaker set that more or less puts broadband providers on the honor system when it comes to indiscriminate handling of your data in transit.

California responded by writing its own law establishing similar (and in some ways expanded) consumer protections. FCC Chairman Ajit Pai, who spearheaded the federal effort to overturn the old rules, was not amused; he called California’s rules “radical,” “anti-consumer,” “illegal” and “burdensome.”

So it was no surprise when, just hours after California governor Jerry Brown signed the bill into law, the FCC filed a lawsuit challenging it.

But the FCC is dealing with a challenge of its own: a lawsuit from a dozen or so internet advocacy companies including Mozilla, Vimeo, Public Knowledge, Etsy and others, alleging all manner of procedural and factual problems with the new federal rules.

If this suit succeeds and the FCC’s new net neutrality rules are rolled back or substantially altered (for instance, the court may find that some section or another is illegal or unenforceable), this could bear on the basis for the agency’s own lawsuit against California. Yes, it’s a bit confusing, and that’s why the state’s attorney general, Xavier Becerra, decided it might be best to wait and not litigate a suit that may be mooted a few months from now.

“We are committed to a free and open Internet for Californians,” AG Becerra said in a statement provided to TechCrunch. “Every step we take, every action we launch is intended to put us in the best position to preserve net neutrality for the 40 million people of our state. We are fighting the Trump Administration’s attempt to repeal net neutrality in the D.C. Circuit Court and we will vigorously defend California’s own net neutrality law.”

Senator Scott Wiener (D-CA) explained in a separate statement that he regrets but understands the necessity of this measure.

“Of course, I very much want to see California’s net neutrality law go into effect immediately, in order to protect access to the internet,” he said. “Yet, I also understand and support the Attorney General’s rationale for allowing the DC Circuit appeal to be resolved before we move forward to defend our net neutrality law in court. After the DC Circuit appeal is resolved, the litigation relating to California’s net neutrality law will then move forward.”

Ajit Pai also issued a statement on the matter, saying he was pleased California was staying implementation of “its onerous Internet regulations.”

“This substantial concession reflects the strength of the case made by the United States earlier this month,” he continued. “It also demonstrates, contrary to the claims of the law’s supporters, that there is no urgent problem that these regulations are needed to address.”

Although the rationale for this delay is understandable, it’s unfortunate that California residents will have to wait months or longer for the protections they supported while this case plays out.

(Updated with statement from California AG Becerra.)

Snapchat is launching its first Mac and Windows software that takes over your webcam and brings its augmented reality effects to other video streaming and calling services. Snap Camera can be selected as a camera output in OBS Skype, YouTube, Google Hangouts, Skype, Zoom and more, plus browser-based apps like Facebook Live so you can browse through Snapchat’s Lens Explorer to try on AR face filters. And through its easily equipped new Twitch extension, streamers can trigger different masks with hotkeys.

You can download the Mac and Windows versions of Snap Camera now. Users can use Lens Explorer to preview effects and see who made them, Star their favorites for easy access and access a tab of your recently used Lenses.

Despite Snap Inc.’s troubles following yesterday’s Q3 earnings announcement that revealed it’d lost 2 million users, causing its share price to hit a new low, Snapchat Camera isn’t about stoking growth. You won’t even have to log in to Snapchat to use it. Instead the goal is to drive more attention to its community AR Lens platform so more developers and creators will make their own effects. “We’re going down the path of providing more distribution channels [for Community Lens creators] and surfacing their work,” Snap’s head of AR Eitan Pilipski tells me. The desktop camera could win Lens creators more attention, and Snapchat connects to the most talented ones to brands for sponsorship deals.

Snapchat first came to the desktop in January with its first embeddable content, designed for newsrooms that wanted to show off citizen journalism on their sites. But now Snapchat content creation is escaping the mobile medium.

Strangely, Snap Camera has no interface of its own. Really, it should have a Photo Booth-style app so you can record photos and videos of yourself with your webcam and share them wherever. “We don’t want to compete in that space. We just want to bring Community Lenses to whatever apps people are using,” Pilipski explains. One major app that won’t support Snap Camera is Apple’s FaceTime. Why? “I don’t know. Apple didn’t comment on that. Believe me we tried,” says Pilipski.

Because there’s “not even a facility to collect the impressions” and users don’t have to log in, Snap won’t be able to add Camera users to its daily active user count. With that number falling from 191 million in Q1 to 188 million in Q2 to 186 million in Q3 as it announced yesterday, Snap really does need more ways to keep people from straying to Instagram Stories. It will have to hope that when video chat users see their friends or family using Snap Camera’s lenses, it will remind them to fire up Snapchat more often. And Lenses could go viral if they show in a Twitch celebrity’s stream.

The Twitch extension comes amidst more announcements at today’s TwitchCon event, including the reveal of Squad Streaming and a karaoke Twitch Sings game for the service’s average of 1 million concurrent viewers and half-million daily streamers.

The Snap camera equips Twitch broadcasters with extra features. They’ll have access to game-themed lenses for League of Legends, PlayerUnknown’s Battlegrounds, World of Warcraft and Overwatch. Viewers will see the QR Snapcode for the Lens on the screen, which they can scan with Snapchat to try the mask on themselves for virality. Streamers get a button that lets viewers subscribe to them, and can set up a “bonus” lens that shows up as a thank you when someone follows them. And with hotkeys, streamers can trigger different lenses, like an angry one for when they lose a game or victory lenses for if they manage to beat all the other Fortnite addicts.

More than 250,000 Community Lenses have been submitted through Snapchat’s Lens Studio since it launched in December, and they’ve been viewed over 1 billion times. Snapchat realized it couldn’t dream up every crazy way people could use AR. Out-Lensing Instagram is critical to Snapchat’s business strategy. The more people that use Snapchat’s AR features, the more the company can charge businesses to promote Sponsored Lenses. With the user count shrinking, Snap needs to show its business is growing to salvage its share price and pull in the outside investment or acquisition it will likely need to make it to profitability. A desktop presence could not only make Snapchat more ubiquitous, but get it in front of older users and advertisers who might be a little scared of its mobile app.

The teens were out in force today in San Jose for the annual TwitchCon game-streaming conference. There, Twitch announced that at any given time, 1 million people are watching it (up from 746,000 last year), and it seemed like many game lovers were at TwitchCon in person to meet some of the nearly half-million web celebs that broadcast each day on the service. Considering Twitch said just 2 million were broadcasting per month in December, the service’s growth is still explosive under Amazon’s ownership.

Amongst the major reveals at TwitchCon were a new Squad Streaming feature that lets up to four people broadcast at once in split-screen that will test with select streamers later this year.

There’s also a new Twitch Sings game built-in partnership with Rock Band-creator Harmonix. Broadcasters can play to perform karaoke (though only with fake versions of songs as Twitch lacks major label music licenses). Viewers can use the chat to request the next song and control the lights on the virtual karaoke stage; broadcasters can sign up here for the Twitch Sings closed beta that starts later in 2018.

And Twitch broadcasters can now use Snapchat’s augmented reality lenses thanks to the new Snap Camera desktop app and accompanying Twitch extension launching today. Streamers can use hotkeys to trigger different Snapchat Lenses, let viewers try those masks by scanning an onscreen Snapchat QR code and reward subscribers with a bonus thank you effect. Read our full story on Snap Camera here.

There were plenty of other minor announcements during the conference’s keynote:

• More than 235,00 streamers now have Affiliate status and are earning money on their channels, while 6,800 have joined its Partnership program so they can earn even more through channel subscriptions and ads.
• Twitch is revamping Gear on Amazon, where streamers can show off products and earn affiliate fees, renaming it Amazon Blacksmith.
• Twitch’s Highlight editor can now stitch together multiple clips from across a broadcasting session.
• New homepage sections will feature up-and-coming streamers, new Partners and Affiliates or streamers local to viewers.
• VIP Badges will let creators recognize their favorite subscribers and moderators.
• Moderators can now see how long someone has been on Twitch, view chat messages that person has sent in the channel and see how many time-outs or bans that account has received in that channel to better understand who to boot.
• 18 billion messages have been sent in Twitch chat and its Whispers feature in 2018, and fans have given creators 85 million Cheers and Subscriptions.
• 150 million Twitch Clips have been created in 2018 to bring the best game stream and other weird content to the rest of the web.
• Twitch users have gifted $9 million worth of subscriptions to fellow users in just 9 weeks.
• Twitch will open its Bounty Board of sponsorship opportunities to 30 more brands, and more Partners and Affiliates in the U.S. and Canada in November.
• The Twitch Rivals in-person gaming tournaments will double to 128 events in 2019. Some will have million-dollar prizes, and it already gave out $5 million in winners’ jackpots last year.

As CEO Emmett Shear made the announcements, audience members hooted and hollered with delight. They out-yelled even Apple’s keynote attendees. Shear shouted out early users who’ve been with it since Twitch was a Y Combinator live-vlogging startup called Justin.tv. “When people have your back and support you for a long time, we think they should be recognized for it,” he said, revealing the new VIP badges and a counter that shows how many months a fan has been a channel’s paying subscriber.

“You spoke and we listened,” Shear said. That truly seemed to be the message of this conference. Facebook’s F8 conferences held in the same San Jose Convention Center often seem to produce updates that are designed to help the company as much as the users. But Twitch has realized it can’t just be useful. It must remain beloved if people are going keep spending 760 million hours per month watching others game, joke and express themselves. Shear concluded, “I think we’re just scratching the surface when it comes to everyone playing together.”

Update: An Interview With Emmett Shear

I spoke with Shear after his keynote to get a sense of Twitch’s priorities and how it’s avoided much of the backlash hitting Facebook, Google, and Twitter. “I don’t think we’re exempt from the problem. We have to work every day on winning the community’s trust. I don’t think you ever get to let your guard down or stop working on that. It’s just through hard work and consistently pushing to build the things that [the streamers] need and that they want.”

Balancing free speech with safety has been a struggle for all the tech platforms, Twitch included. “I think this is the issue of our time. This is the thing that every tech, media, and communications company in the world has to grapple with. We’re not shy about asking people who don’t abide by our community standards to leave” Shear tells me.

I asked whether he’d kick Alex Jones off the platform if he joined, even before violating Twitch’s own rules due to his behavior elsewhere. “We don’t talk about individual cases, generally speaking. Trying to police anyone’s behavior across the internet is hard because of…the internet not being able to tell you’re a dog” he says, referring to the old adage about anonymity on the web. “But we believe for example that harassment on another platform, it’s still you. We have to be able to know it verifiably is you. You can’t jump to conclusions. But if it is verifiably you and you’ve gone off Twitch to harass people, we have no problem banning you for that behavior.”

As for the competitive landscape, Shear beamed “I think it’s awesome to see such vigorous invest in livestreaming globally. I’ve been working on livestreaming since 2006. It’s nice to get the validation that everyone realizes it’s a good idea too…a decade later.” Shear is believed to be under a five-year vesting schedule at Amazon that’s set to complete next year. “I’ve felt incredibly autonomous and supported by Amazon” he tells me. But is he going to leave? “You never know what the future holds. I’m loving my job. I’m loving to getting to work on Twitch, and the people that I work with. Being part of Amazon is pretty good. Compared to friends I’ve talked to raising money from VCs, I think I prefer the current setup.”

Tesla, which reported its first quarterly profits in two years Wednesday, is looking to extend its earnings streak by bringing its new Model 3 to customers beyond North America. And part of that plan involves accelerating its manufacturing plans in China.

Tesla saw its revenue skyrocket to $6.8 billion in the third quarter (and a $312 million profit) thanks to sales of its new Model 3 vehicle, despite production bottlenecks and more recent issues with delivery logistics. The company was able to achieve that profitability milestone just through sales in the U.S. and Canada. That leaves two other massive markets on the table. Cue Europe and China.

Tesla said Wednesday it will start to take orders for the Model 3 in Europe and China before the end of 2018. Tesla said it will begin deliveries of the Model 3 to Europe early next year.

“The mid-sized premium sedan market in Europe is more than twice as big as the same segment in the U.S.,” Tesla said in its shareholder letter released Wednesday. “This is why we are excited to bring Model 3 to Europe early next year.”

Notably, the company is further accelerating its timeline for China and said it will bring portions of Model 3 production to the country next year.

“We are aiming to bring portions of Model 3 production to China during 2019 and to progressively increase the level of localization through local sourcing and manufacturing,” Tesla said in its earnings report. “Production in China will be designated only for local customers.”

Tesla said earlier this month it plans for as rapid build out of a factory in China. But there’s something new here. The term “portions of Model 3 production” is the important phrase. This could be referring to a term used in the manufacturing world known as a complete knock down. CKD is basically a kit of non-assembled parts of a product, like say a Model 3. It’s a strategy used to avoid tariffs when shipping to foreign countries.

Tesla has plans to build a factory in Shanghai, but construction hasn’t even begun yet.

The company secured in October rights to about 210 acres of land in Lingang, Shanghai, the site of the electric automaker’s planned factory and its first outside of the U.S.

Tesla warned in its production and delivery report in early October that tariffs, combined with the cost of shipping its vehicles via ocean carrier and the lack of access to cash incentives available to locally produced electric vehicles, has put the company at a disadvantage in China. Tesla reiterated those cost constraints in its third-quarter earnings report.

Tesla reached a deal in July with the Shanghai government to build a factory that it says will be capable of producing 500,000 electric vehicles a year. Once construction begins, it will take about two years until Tesla can produce vehicles. It will be another “two to three years before the factory is fully ramped up to produce around 500,000 vehicles per year for Chinese customers,” a Tesla spokesman said at the time.

Facebook announced today that it has removed 8.7 million pieces of content last quarter that violated its rules against child exploitation, thanks to new technology. The new AI and machine learning tech, which was developed and implemented over the past year by the company, removed 99 percent of those posts before anyone reported them, said Antigone Davis, Facebook’s global head of safety, in a blog post.

The new technology examines posts for child nudity and other exploitative content when they are uploaded and, if necessary, photos and accounts are reported to the National Center for Missing and Exploited Children. Facebook had already been using photo-matching technology to compare newly uploaded photos with known images of child exploitation and revenge porn, but the new tools are meant to prevent previously unidentified content from being disseminated through its platform.

The technology isn’t perfect, with many parents complaining that innocuous photos of their kids have been removed. Davis addressed this in her post, writing that in order to “avoid even the potential for abuse, we take action on nonsexual content as well, like seemingly benign photos of children in the bath” and that this “comprehensive approach” is one reason Facebook removed as much content as it did last quarter.

But Facebook’s moderation technology is by no means perfect and many people believe it is not comprehensive or accurate enough. In addition to family snapshots, it’s also been criticized for removing content like the iconic 1972 photo of Phan Thi Kim Phuc, known as the “Napalm Girl,” fleeing naked after suffering third-degree burns in a South Vietnamese napalm attack on her village, a decision COO Sheryl Sandberg apologized for.

Last year, the company’s moderation policies were also criticized by the United Kingdom’s National Society for the Prevention of Cruelty to Children, which called for social media companies to be subject to independent moderation and fines for non-compliance. The launch of Facebook Live has also at times overwhelmed the platform and its moderators (software and human), with videos of sexual assaults, suicides, and murder—including that of an 11-month-old baby by her father—being broadcast.

Moderating social media content, however, is one noteworthy example of how AI-based automation can benefit human workers. Last month, Selena Scola, a former Facebook content moderator, sued the company claiming that screening thousands of violent images had caused her to develop post-traumatic stress disorder. Other moderators, many of whom are contractors, have also spoken of the job’s psychological toll and said Facebook does not offer enough training, support, or financial compensation.

Facebook is working on a new app that it hopes could win back the attention of teens while capitalizing on its recently-attained major label music licenses. Facebook is building a standalone product where users can record and share videos of themselves lip syncing or dancing to popular songs, according to information from current and former Facebook employees.

The app is designed to be a standalone competitor to Musically, which was a hit with teens and even pre-teens before the 60 million monthly user product was acquired by Chinese tech giant ByteDance for around $1 billion and rolled into the company’s TikTok app.

“It’s basically TikTok/Musically . It’s full-screen, built for teens, fun and funny, and focused on creation” one source told me. “A lot of what they’re doing is just trying to be cool and ment for this report.

The product is being built by members of Facebook’s video and Watch team under leadership from Facebook’s principal lead product designer Brady Voss, a source says.

Voss previously worked on Facebook’s TV app as well as the recently shut down Hello standalone app. For a 2016 hackathon project he presented to Mark Zuckerberg, Voss made a technology called Montage that would stitch together photos of say a snowboarder doing a trick into a single image like a still timelapse. Now he’s back at making creative expression simpler with Lasso. “Brady is great with fun new camera and video things” a source said.

Facebook has been investigating the teen music app space since 2016, when a source says the company spun up a research project to look into Musically. There were suspicions that Musically might not be as popular as it touted, and Facebook eased off.

Then earlier this year, Facebook’s music efforts were reinvigorated when it secured licensing deals with all the major record labels. At first, this just kept users’ videos, including Musically-style lip syncing clips, from being taken down for copyright infringement.

But soon it launched music stickers on Instagram that let you add soundtracks to your Stories, and that feature rolled out on Facebook today. The company also began experimenting with a Lip Sync Live feature for livestreaming karaoke, and today Facebook opened it for Pages and began showing some songs’ lyrics on screen. It plans to soon allow users to pin their favorite songs to their profile so friends can listen to a segment in what feels like a throwback to MySpace Music.

Meanwhile, TikTok is on the rise. The app has climbed from the #32 overall iPhone app in the US three months ago to reach #5 today, according to App Annie.

Now just 5 percent of US teens cite Facebook as their favorite social platform, according to a Piper Jaffray survey. The percentage who use it monthly has dropped from 60 percent to 36 percent since Spring 2016. Facebook needs new ways to engage teens beyond Instagram and WhatsApp, and a standalone music app potentially devoid of its own branding could be a better approach than cramming teen features like Lip Sync Live into its uncool main app.

Facebook already tried and failed to win back the youth with standalone apps like Poke, Slingshot, Bolt, Flash, and other variations on Snapchat. But with US giants like Snap and YouTube neglecting to build proper tools for video music creation, Facebook has a shot to challenge China’s ByteDance . Most people aren’t interesting on camera, especially awkward teens. But with the right soundtrack,  a stupid selfie video can become epic, or at least silly enough to watch.