Last Week in AI #133: Deepfake cyberattacks, Project Maven updates, AI's geographic concentration in the U.S.

Reports of deepfakes used in phishing and biometric attacks, Pentagon's Project Maven finds new partners after Google, and concentrated AI development risks winner-takes-most dynamics

Top News

Deepfakes in cyberattacks aren’t coming. They’re already here. 

Concerns about the potential for deepfakes and synthetic media to cause harm still abound, and they are now being used in cyberattacks. In a dark parody of open-source tutorials on novel technologies, threat actors are using the dark web to share best practices and methods for using deepfakes to compromise organizations. The concerns are not just theoretical: Deepfakes are being used in phishing attempts and to fool biometric-based authentication. As these technologies become more capable, the threat landscape will likely become worse.

Project Maven: Amazon And Microsoft Scored $50 Million In Pentagon Surveillance Contracts After Google Quit

In 2018, Google got a large amount of backlash from its employees for accepting a contract from the Department of Defense to develop AI technology for its drones, named Project Maven. Google ultimately cancelled this commitment, but new analysis shows that Amazon and Microsoft accepted contracts with $50 million that have similar goals. Both Microsoft and Amazon have likewise seen employee backlash over past involvement with government contracts. It’s not clear if the companies are still working on these projects, and they have not commented on this news. 

In the US, the AI Industry Risks Becoming Winner-Take-Most

Brookings Institution released a new report on how cities in the U.S. approach AI development and found that research, talent, and funding are highly concentrated in a few metro areas. In particular, and perhaps unsurprisingly, San Francisco Bay Area has four times higher AI activity than other top cities. The report warns that such concentration may lead to groupthink and more winner-take-most dynamics, and it encourages using federal policies to diversify AI investments in different locations. Such policies may include expanding local educational opportunities, setting tax breaks for local AI companies, and awarding government contracts based on development targets. Lastly, the study identified many metro areas that have high levels of AI research activities but not economic activities. These areas should be prioritized for business development to avoid graduates and professors moving elsewhere to seek similar opportunities.

Share

Check out our weekly podcast covering these stories! 
Website | RSS | iTunes | Spotify | YouTube

Other News

Research

Greedy AI Agents Learn to Cooperate - "Imagine you're sitting at a casino's poker table. Someone has explained the basic rules to you, but you've never played before and don't know even the simplest strategies."

Sharing learnings from the first algorithmic bias bounty challenge - "In this post, we’ll share what we learned through creating and hosting this challenge, what the submissions taught us, and what’s next. "

A New Google AI Research Study Discovers Anomalous Data Using Self Supervised Learning - "Anomaly detection is one of the most common machine learning applications in various areas, from industrial defect identification to fraudulent financial detection. One-class classification is beneficial for anomaly detection."

IBM Leverages Reinforcement Learning to Achieve SOTA Performance on Text and Knowledge Base Generation - "Knowledge bases (KBs) can be used to store complex structured and unstructured information, and are a powerful tool for capturing real-world information with complex relationships."

DeepMind aims to marry deep learning and classic algorithms - "Now DeepMind has set its sights on another grand challenge: bridging the worlds of deep learning and classical computer science to enable deep learning to do everything."

Columbia to Launch $25 Million AI-based Climate Modeling Center - "Funded by the National Science Foundation, the center will leverage big data and machine learning to improve climate projections and motivate societies to invest in policies and infrastructure to confront rising seas and warmer temperatures."

Applications

Astronauts in space will soon resurrect an AI robot friend called CIMON - "The robot, called CIMON-2 (it's short for Crew Interactive Mobile Companion) worked alongside two European astronauts on past missions to the station in recent years and just got a software upgrade that will enable it to perform more complex tasks with a new human crewmate later this year."

Machine learning technique detects phishing sites based on markup visualization - "Machine learning models trained on the visual representation of website code can help improve the accuracy and speed of detecting phishing websites. This is according to a paper (PDF) by security researchers at the University of Plymouth and the University of Portsmouth, UK."

New Algorithm Can Identify Pre-Alzheimer's Brain Changes With Over 99% Accuracy - "One of the most important ways in which artificial intelligence algorithms are proving beneficial is in diagnosing disease much faster than mere human beings can – and a new system for detecting potential indicators of Alzheimer's has reached close to 100 percent accuracy."

iRobot’s newest Roomba uses AI to avoid dog poop - "Finally, decades of research into artificial intelligence are paying off"

Business

What Databricks’s $1.6 billion funding round means for the enterprise AI market - "The latest winner of the growing interest in enterprise AI is Databricks, a startup that has just secured $1.6 billion in series H funding at an insane valuation of $38 billion. This latest round of investment comes only months after Databricks raised another $1 billion."

Ex-Googlers raise $40 million to democratize natural-language AI - "Their startup, called Cohere, is backed by AI luminaries Geoffrey Hinton and Fei-Fei Li. 5 minute ReadThe ability of computers to understand and generate language took a huge leap forward in 2017 when researchers at Google developed new natural language AI models called Transformers."

VW will deploy Argo AI-powered autonomous ID.Buzz vans - "Volkswagen announced it will begin testing self-driving, ID.Buzz-based electric shuttles in 2025. Powered by Argo AI, a Pittsburgh-based company the firm holds a stake in, the vans will begin carrying users in Hamburg."

IDC Forecasts Companies to Spend Almost $342 Billion on AI Solutions in 2021 - "Worldwide revenues for the artificial intelligence (AI) market, including software, hardware, and services, is estimated to grow 15.2% year over year in 2021 to $341."

WeRide unveils China’s first Level 4 self-driving cargo van,WeRide Robovan - "WeRide, a world-leading autonomous driving company, forayed into urban logistics industry with the unveiling of WeRide Robovan, the very first L4 self-driving cargo van in China."

Gartner: AI is moving fast and will be ready for prime time sooner than you think - "Companies have two to three years to lay the groundwork for successful use of generative AI, synthetic data and orchestration platforms."

Alibaba-Backed AI Startup Megvii Receives Go-Ahead for 6 Billion Yuan Shanghai IPO - "On September 9, an official announcement issued by the Shanghai Stock Exchange showed that the initial listing of AI firm Megvii Technology on the Science and Technology Innovation Board has been approved."

Concerns

New AI system fills rifle sights with extensive, easy-to-digest info - "The system has AI powering its applications and can perform functions including ballistic correction so that the reticle is in the right position."

Why are so many scummy, scammy AI companies thriving? - "$urely, there'$ a $imple rea$on, let'$ $ee if we can figure it out."

Facebook is very sorry that we keep noticing its racist AI - "Making AI that isn't racist is too hard and not using racist AI is apparently out of the question"

Do we need humans for that job? Automation booms after COVID - "Faced with worker shortages and higher labor costs, companies are starting to automate service sector jobs that economists once considered safe, assuming that machines couldn’t easily provide the human contact they believed customers would demand."

Money, mimicry and mind control: Big Tech slams ethics brakes on AI - "All these technologies were curbed by panels of executives or other leaders, according to interviews with AI ethics chiefs at the three U.S. technology giants."

How to Opt Out of Facial Recognition at the Airport - "Facial recognition technology is becoming more widespread, but opting out isn't always obvious."

Analysis

Six Essential Elements Of A Responsible AI Model - "New ethical and moral questions continue to emerge as we expand how we use artificial intelligence in business and government."

Could Robots From Boston Dynamics Beat Me in a Fight? - "A parkour showcase of the agile humanoids is both wondrous and unsettling."

A.I. Can Now Write Its Own Computer Code. That’s Good News for Humans. - "As soon as Tom Smith got his hands on Codex — a new artificial intelligence technology that writes its own computer programs — he gave it a job interview."

Deepfake tech doesn’t have to be ‘bad’ - "It's a fine line between funny and horrific"

Explainers

A friendly introduction to machine learning compilers and optimizers - "I have a confession to make. I cried during the compiler class in college. I became a machine learning engineer so that I wouldn’t have to worry about compilers. However, as I learned more about bringing ML models into production, the topic of compilers kept coming up."

Copyright © 2021 Skynet Today, All rights reserved.