Site iconSite icon ForkLog

Algorithms used to track criminals; Egypt detains spy-robot and other AI news

Algorithms used to track criminals; Egypt detains spy-robot and other AI news

We aim to inform readers not only about developments in the bitcoin industry, but also about what is happening in adjacent tech spheres — cyber security, and now the world of artificial intelligence (AI).

ForkLog has gathered the week’s most important AI news.

  • Russia unveiled 36 of 217 AI standards.
  • The Moscow Metro launched a fare-payment service using facial recognition at all stations.
  • Experts criticized Apple’s and the EU’s plans to scan users’ devices for prohibited materials.
  • In Scottish schools, a facial-recognition system for children to pay for meals will be installed.
  • Apple unveiled processors for laptops with an integrated neural chip.
  • In Egypt, a spy-robot Ai-Da was detained on suspicion of espionage.
  • The week’s top AI deals.

Russia Unveils National AI Standards

The Technical Committee for Standardization 164 \”Artificial Intelligence\” presented 36 national AI standard projects.

They were developed in the following areas:

A total of 217 AI standards are planned by 2024.

The Moscow Metro has rolled out face-pay at all stations

Since 15 October 2021, the Face Pay system for facial-recognition fare payment went live at all Moscow Metro stations.

Turnstile with biometric identification. Data: Moscow City Hall.

According to city authorities, passenger data is securely encrypted and stored in data centres accessible only to Ministry of Internal Affairs staff. They added that biometrics are collected solely at the user’s request and are not intended to be shared with the police.

Representatives of the digital rights group Roskomsvoboda say it is a dangerous step toward population control, and call for full transparency about how the algorithms operate.

Russian lawmakers draft bill to regulate recommendation systems

The State Duma will consider a bill to regulate recommender systems.

Under the document, social networks, news aggregators and audiovisual services would fall under its scope. Authorities would require platforms to warn users about the data used and to allow disabling personalization algorithms.

Lawmakers say that recommender systems can create abnormal consumer interest, influence public opinion and raise risks of social conflicts. Business owners fear that the bill will slow the development of technologies and hurt financial results.

The Investigative Committee of Russia uses AI to locate serial offenders

The Investigative Committee of Russia launched an AI-based program for constructing portraits of serial criminals. It operates with up to 96% accuracy.

The system estimates the suspect’s approximate age, presence or absence of mental illness, criminal history, family, personal transport, and likely residential area. This information, along with other case details, helps investigators conduct investigations more effectively.

The technology is already used to solve crimes.

Experts criticise Apple and EU plans to scan devices for illegal materials

More than 40 cybersecurity experts criticized Apple and the European Union’s plans to scan devices for illegal material.

They published a 46-page document in which they called the technology “dangerous, invasive and ineffective.” They argue that the tech giant’s plans to detect images of child abuse, and the EU’s aims to identify CSAM and terrorism on encrypted devices, could enable government surveillance.

Apple declined to comment on the report. There was no word from the EU.

Study: Twitter algorithms promote right-wing content more often than left-wing

Twitter’s algorithms favour posts from right-wing ideologues in six of seven countries, rather than left-wing ones, according to a study by company researchers.

The social-media team considered two questions: whether the models amplify political content from elected officials and whether some groups saw greater post reach. They used tweets from news agencies and some politicians from Canada, France, Germany, Japan, Spain, the United Kingdom and the United States.

The company does not know why this is the case. They suggest that algorithmic bias toward right-wing ideas may reflect different party strategies for promoting content on social media.

Experts note that the question requires further study.

In Scottish schools, facial recognition for children’s lunches will be deployed

Nine Scottish schools will apply facial recognition for lunch payments. The move was supported by 97% of parents and children.

The technology is intended to minimise touch during the pandemic, but is mainly aimed at speeding up transaction times.

CRB Cunningham will install the system. It does not use real-time biometric authentication on its devices; instead, it compares encrypted ‘faceprint’ templates.

Apple unveils M1 Pro and M1 Max processors with a dedicated Neural Engine

Apple announced 5-nanometer single-chip ARM-based M1 Pro and M1 Max. They are equipped with a 16-core Neural Engine.

According to company representatives, the Neural Engine is used to improve image quality, obtain sharper video and more natural skin tones on the built-in camera. The chip also enables faster on-device ML processing.

Specifications of the M1 Pro and M1 Max processors. Data: Apple.

\n\n\n\n

The M1 Pro has 33.7 billion transistors. It features ten compute cores, eight high-performance and two efficiency cores, and 16 graphics cores with 2,048 execution units. The chip supports up to 32 GB of unified memory.

\n\n\n\n

The M1 Max has 57 billion transistors. It features ten compute cores and 32 graphics cores with 4,096 execution units. The processor supports up to 64 GB of unified memory.

\n\n\n\n

According to Apple representatives, both chips deliver a 1.7x improvement in performance per watt versus the M1 chip.

\n\n\n\n

DeepMind buys MuJoCo, the robot-simulation developer

UK AI lab DeepMind, a subsidiary of Alphabet, acquired MuJoCo, the company that develops a simulation environment for training robots and calculating their movement physics.

MuJoCo robot-learning environment. Data: DeepMind.

The company intends to use the platform as a virtual lab for modelling various processors.

DeepMind also made the simulator free and open to the public.

The price of the acquisition, as well as terms of the deal, were not disclosed.

In Egypt, a robot was detained on suspicion of espionage

Egyptian authorities detained Ai-Da, the artist-robot, on suspicion of espionage.

The device was due to participate in the Forever is Now exhibition near the pyramids at Giza. However, customs officers detained the robot and its developer for 10 days over security concerns, prompted by the cameras and modem on board.

Ai-Da robot, detained by Egyptian authorities over alleged espionage. Data: The Guardian.

In the evening of 20 October 2021 the suspects were released and flown to Cairo.

A free AI-powered tool to remove unwanted objects from photos

The ClipDrop engineering team developed the web tool CleanUp.pictures for removing unwanted objects from images.

It is based on the open-source LaMa neural network, which automatically inpaints the selected areas.

Anyone can use the tool. We at ForkLog AI tested it, and here is what we got.

Result of the tool. Data: CleanUp.pictures.

Top AI deals of the past week

From 17 to 23 October 2021, AI startups attracted over $646 million. Here are the most notable deals.

Also on ForkLog:

Subscribe to ForkLog’s AI news on Telegram: ForkLog AI — all the AI news!

Exit mobile version