Deepfakes were rated as the most concerning AI crime in a new UCL report. The report explored 20 ways that AI technology could result in crime and terrorism over the next 15 years. The most concerning uses, after deepfakes, were driverless cars as weapons, spear phishing, AI-synthesised fake news, disrupting AI-controlled systems, and harvesting data for large-scale blackmail.

More:

• The study, published in Crime Science, was funded by the Dawes Center for Future Crime at UCL. 31 experts from academia, defense, and law enforcement ranked the crimes based on their potential for harm, criminal gain, ease of deployment, and difficulty to stop.
• Deepfakes have the potential to discredit public figures, extract money through impersonations, and inflict other harm, which could result in widespread distrust of audio and visual media, the report noted.
• AI crimes ranked as less harmful included autonomous attack drones, AI-assisted stalking, sales of fraudulent services, and misuse of military robots.
• Professor Shane Johnson, director of the Dawes Center, said its the first report in a series aimed at identifying crime threats linked to emerging technologies.
• Earlier this week, a deepfake video of Nancy Pelosi that makes her appear intoxicated went viral on Facebook for a second time.
• Mashable compiled a list of 13 deepfakes that will "seriously mess with your brain."

Portland, Maine, became the 13th U.S. city to ban its police and other city officials from using facial recognition, making it the first city to do so outside of Massachusetts and California. After months of delays, Portland councilors on Monday voted unanimously in favor of the ban, which was supported by the ACLU and members of the Black Lives Matter movement.

More:

• Councilor Pious Ali said he proposed the ban to prevent racial profiling.
• Prior to the vote, councilors removed an enforcement provision targeting city workers who violate the ban, which weakened it.
• Some city officials were concerned the ban would prevent Portland's airport from someday using facial recognition to screen international travelers.
• The Portland Police Department and other city departments said they don't currently use the technology.
• San Francisco, Oakland, and Boston are among the other cities that have outlawed municipal use of facial recognition.

Related:

• Fawkes - a software that helps people protect their online images from unauthorized facial recognition - will soon get a free app version.
• Facial recognition algorithms from Panasonic, Tencent, and other companies failed up to 50% of the time when tested on faces wearing digitally-applied masks, a new government study shows.
• Rite Aid recently stopped using facial recognition after quietly installing the software in 200 of its U.S. drug stores, many of which were in predominantly Black or Latino neighborhoods.
• Facebook agreed to settle a facial recognition lawsuit for $650m, one of the largest payouts ever related to data privacy.

Worldwide AI revenues are forecasted to rise to $156.5b in 2020, a 12.3% increase year over year, according to a new forecast. The largest AI technology group is software, which makes up 80% of all revenues in the AI market.

More:

• The 12.3% rise is slightly slower than in prior years, likely due to the pandemic. Investments in AI will quickly recover, however...

Amazon removed its Alexa Guard feature - which listens for burglar alarms and other signs of forced entry - for its U.K. customers. Each customer in the country that enabled the feature was also sent a £10 gift card.

More:

• Reports didn't specify why Amazon U.K. rolled out the feature, which is apparently still available in the U.S.
• An Amazon email called it a "technical issue" and specifies that it "should not have been possible for our customers in the U.K. to sign up." The Amazon gift card was sent as a "token of appreciation" to customers.
• Once the Echo-enabled Guard feature hears signs of a potential break-in, it alerts homeowners in their Alexa app.
• Forbes' Barry Collins speculates that Amazon can't launch Guard in the U.K. because it breaches intellectual property or violates the EU's stricter data privacy laws.

New job postings in AI/ML:

Data Engineer, Machine Learning Infrastructure & Analytics
Argo AI | Pittsburgh, Pa. and and Palo Alto, Calif.

Software Engineer, Machine Learning
Grammarly | Vancouver

AI/ML - Software Engineer (Siri Rotational Program)
Apple | Seattle...

Quick Hits:

• In a new Deloitte Consulting survey, 83% of respondents said AI and machine learning are the most likely technologies to transform health care over the next decade.
• Royal Bank of Canada debuted a private AI cloud that can analyze more than 13 million client records in less than 20 minutes.
• A quantum version of the Chinese board game Go could prove a tough challenge for AI systems, New Scientist reports.
• Golf brand Callaway used AI and machine learning to design a new range of clubs.
• A new study in Genome Biology confirms that deep learning on biological networks is both feasible and practically useful.
• TU Braunschweig researchers submitted a paper that reviews the staging and preventing of adversarial image-scaling attacks against ML systems.
• Observe.AI, a contact center AI firm, has hired Rich Wang as VP of Customer Success.
• A Redditor shared an animation of a neural network training with TensorFlow.
• Men's Journal says this is the best-customized skincare for men.*

*This is a sponsored post.

Tweet of the Day: Johns Hopkins cognitive scientist Chaz Firestone shares a funny interaction with GPT-3.