AI

Dangerous Deepfakes

A couple weeks ago, cyber firm Sensity revealed that a pornographic “deepfake ecosystem” has spread across the messaging app Telegram. Its researchers discovered that bots have created over 104,000 fake, AI-generated nude pictures of real women. 

• Images were shared in private or public channels beyond Telegram, for “public shaming or extortion-based attacks.” The bot network was boosted on VK, Russia’s largest social media network.

What’s going on?

As we frequently write, deepfake technology has been commoditized. 

“Progress in synthetic media has been phenomenal in the last few years,” Sensity CEO Giorgio Patrini told Emerging Tech Brew, “2020 is definitely the tipping point for these technologies to become pervasive, for the good but also for the bad, unfortunately.”

• A few years ago, someone may have needed AI training to generate deepfakes. Now, they can leverage off-the-shelf models for malicious or harmless purposes. 

Counter-measures

Telegram was unresponsive when Sensity reached out with its findings.

Most countries don’t have clear laws governing the use of deepfakes, and instead resort to existing privacy or anti-harassment laws that are already on the books. “More legislation could help,” Patrini said. 

• Virginia has imposed criminal penalties on disributing nonconsensual deepfake pornography. California has passed a law that outlaws dissemination of deepfakes featuring politicians within 60 days of an election.

Detection 

Sensity blends “automated detection of visual threats”—via web crawlers and computer vision—with the work of threat intelligence experts, who search for malicious activity among underground communities on the dark web. 

Private groups and encrypted messaging apps pose a challenge to effective deepfake detection (and content moderation writ large). In this case, Sensity uncovered activity in public channels by searching for keywords on Telegram. 

• “Since those services are not ‘illegal,’ or perhaps live in a grey area of the law,” Patrini reasons, the bot network’s operators can freely promote and monetize their malicious service. 

Zoom out: “I strongly believe that most [platforms or messaging companies] won’t be developing internally the needed layer of security for deepfakes,” Patrini said. “It’s not their business, at the end of the day.”

ROBOTICS

Tech We Ship

Online shopping to fill the void? You’re not alone: The average consumer spent $254 on impulse buys in the past month. 

And since there are only a handful of companies responsible for getting yet another pair of sweatpants from your phone screen to your door, the pressure is on ahead of what’s projected to be one of the largest-ever holiday shopping seasons. 

On Thursday, FedEx detailed its tech game plan for getting through it.

Robotic arms: FedEx’s Memphis facility installed four robotic arms in March, which move packages from collection bins to conveyor belts. 

• Randall, Colin, Sue, and Bobby can each sort ~1,300 packages per hour. 

Delivery robots: Roxo, the autonomous bot FedEx began testing last year, can make local same-day deliveries. What’s new this year? FedEx Autonomous Mobility Ecosystem (FAME): A management platform—for use with Roxo and future autonomous tools—that can dispatch a bot for a particular task or control a fleet. 

Autonomous vehicles: FedEx Ground is testing out autonomous vehicles to help move trailers from loading docks. FedEx told us it’s also invested in ~5x more of Vecna Robotics’s autonomous “tuggers,” which use sensors to navigate and move bulky packages between loading spots. 

Looking ahead: If you thought we were already in an industrial robotics boom...just wait. 

SPONSORED BY VOICE

If You’re a Human, Keep Reading

While social media is great for sharing photos from Uncle Steve’s epic dinosaur-themed wedding (very cool), it’s also filled with fake human bots that share information that confuses and divides us (very uncool). 

Which is why Voice is building a human-first social media platform. It’s what social media should be—free of robo-mobs, fake accounts, catfish, and shady algorithms. 

Voice is currently invite-only, but Emerging Tech Brew readers can get early access to the future of social media. 

So if you’re not a bot, sign up for Voice today.

AI

How can You Tell if an Algorithm is Biased?

On Friday, we published a standalone piece that explored the world of algorithmic auditing. In case you missed it then, here’s a quick overview. 

What it is: Algorithmic auditing is a new but quickly developing field that aims to suss out bias in code before it causes harm in the real world. 

Why it matters: “Real people’s lives [hang] in the balance” of how algorithms are used, says Kush Varshney, distinguished research staff member at IBM Research. 

These systems are often shielded by “corporate secrecy laws,” says Amba Kak, director of global policy at the AI Now Institute. 

• That’s why—although algorithmic audits can be internal or external, voluntary or imposed—the "most impactful, in terms of creating systemic change" have been conducted by independent research and advocacy groups, she says. 

For an in-depth reveal on the process...click here to read the full story from Hayden.

BITS & BYTES

Stat: Ant Group, the Chinese fintech unicorn and Alibaba affiliate, is expected to raise up to $34.4 billion Thursday in the world’s largest stock market debut.

Quote: “It’s my party trick: where you can tell me any job...and I can find a way that either AI is already impacting it or will in the next three to five years.” —Tabitha Goldstaub, chair of the UK’s AI Council, in an interview about her new book: How To Talk To Robots 

Read: The Verge dove into Waymo’s in-depth report on self-driving operations in Phoenix. 

SPONSORED BY MISO

Invest in the new fiscal beef chief. There’s a robot named Flippy who cooks burgers, and that’s the least interesting part of this story. The most interesting part? You can invest in this cash cow (flipping) robot right now. This tech is the future of the Quick Service Restaurant industry. The campaign ends this month, so invest today

WHAT ELSE IS BREWING

• Huawei is planning to build its own chip plant in Shanghai, per the FT. 
• Daimler invested in Luminar, as the lidar maker prepares to go public via SPAC. 
• A U.S. judge blocked the Commerce Department from banning TikTok.
• Harley Davidson spun out Serial 1 Cycle Company, an e-bike company. 
• Tencent says its Honor of Kings game has surpassed 100 million monthly active users. 
• Snapchat introduced full-body tracking AR lenses for Halloween. 

WHAT'S BREWING THIS WEEK

Monday: PayPal earnings. 

Tuesday: Election Day (Vote.org has a polling place locator). 

Wednesday: Earnings (Qualcomm, Match); IoT Tech Expo runs through Friday; National Stress Awareness Day (fitting). 

Thursday: Ant Group’s expected IPO; Earnings (Alibaba, Uber, Roku, Nintendo, EA, Take-Two Interactive, Square, Dropbox, Cloudflare, GM, Zillow). 

Friday: UW’s virtual conference on autonomous operations in space. 

TECH THINGAMABOBS

For (virtual) cocktail conversation: Verizon used millions of Smithsonian artifacts to create AR museum experiences. Choose from exhibitions on air and space, innovative inventions, and more. 

For an all-in-one keyboard: Check out the new $70 Raspberry Pi 400, a computer that fits within a keyboard. According to its creators, the new Pi is ~40x more powerful than original Pi and offers performance “indistinguishable from a legacy PC for the majority of users.” 

ICYMI

Catch up on the top Emerging Tech Brew stories from the past few editions: 

• Search tech: Apple vs. Google 
• China’s new five-year plan
• How Google Assistant learns slang