It's been estimated that China's government has detained as many as a million members of the country's Muslim population in so-called "re-education camps," in part of a campaign that has alarmed human rights activists across the world. This week, drawing on 403 pages of leaked government documents, The New York Times published new details of how the ongoing crackdown took shape under Chinese President Xi Jinping and other leadership in the Communist Party of China, how government workers who resisted the plan were sidelined, and what officials were instructed to tell young people whose families had been detained. "They're in a training school set up by the government to undergo collective systematic training, study and instruction," the talking points read, adding, "You have nothing to worry about." The Chinese government's campaign against those it says have been exposed to extremism is centered on an autonomous region, Xinjiang, where nearly half of the 25 million residents are a Muslim people called the Uighurs. Earlier in November, a FRONTLINE documentary called In the Age of AI examined how, as part of its crackdown involving the Uighurs, China's government has made Xinjiang a test project for forms of extreme digital surveillance.

Think lawsuits involving humans are tricky? Try taking an intelligent robot to court. While autonomous robots with humanlike, all-encompassing capabilities are still decades away, European lawmakers, legal experts and manufacturers are already locked in a high-stakes debate about their legal status: whether it's these machines or human beings who should bear ultimate responsibility for their actions. The battle goes back to a paragraph of text, buried deep in a European Parliament report from early 2017, which suggests that self-learning robots could be granted "electronic personalities." Such a status could allow robots to be insured individually and be held liable for damages if they go rogue and start hurting people or damaging property.

Chinese companies have exported artificial intelligence surveillance technology to more than 60 countries including Iran, Myanmar, Venezuela, Zimbabwe and others with dismal human rights records, according to a report by a U.S. think tank. With the technology involving facial recognition systems that the Communist Party uses to crack down on Uighurs and other Muslim minorities in China's far western Xinjiang region, the report calls Beijing a global driver of "authoritarian tech." The Carnegie Endowment for International Peace released the report amid concerns that authoritarian regimes would use the technology to boost their power and data could be sent back to China. "Technology linked to Chinese companies -- particularly Huawei, Hikvision, Dahua and ZTE -- supply AI surveillance technology in 63 countries, 36 of which have signed onto China's Belt and Road Initiative," it said. Critics say the BRI, President Xi Jinping's signature cross-border infrastructure project, is intended to draw countries in Asia, Africa and Europe deeper into Beijing's economic orbit.

California's passage of their "GDPR-lite" caught people off guard. We think this is part of a trend we've studied for a long time. Much of the current analysis misses key points, so it seems worth explaining. About two years ago, we asked several thought leaders in the U.S. about the odds we'd see legislation like the E.U. GDPR provides clear rights to E.U citizens, controlling data captured on-line.

Deepfakes are spreading fast, and while some have playful intentions, others can cause serious harm. We stepped inside this deceptive new world to see what experts are doing to catch this altered content. Chances are you've seen a deepfake; Donald Trump, Barack Obama, and Mark Zuckerberg have all been targets of the computer-generated replications. A deepfake is a video or an audio clip where deep learning models create versions of people saying and doing things that have never actually happened. A good deepfake can chip away at our ability to discern fact from fiction, testing whether seeing is really believing.

Key point: Faster, more capable processors will enable the aircraft's avionics, radar, sensors and communications technologies to better identify and attack enemy targets. Faster computer processors, AI-infused algorithms able to merge or "fuse" sensor information and automated maintenance and checklists are informing emerging pilot tactics aimed at anticipating future threat environments. Various applications of AI now perform a wide range of functions not purely restricted to conventional notions of IT or cyberspace; computer algorithms are increasingly able to almost instantaneously access vast pools of data, compare and organize information and perform automated procedural and analytical functions for human decision-makers. When high-volume, redundant tasks are performed through computer automation, humans are freed up to expend energy pursuing a wider range of interpretive or conceptual work. "The bottom line is the next big thing that is going to enable the US to maintain its qualitative edge is the seamless and ubiquitous sharing of information.

Emphasising on the growth of the Indian startup ecosystem, NITI Aayog chief Amitabh Kant said that across different sectors, India has gone digital. "As a result, India has become hugely data-rich, even before it has become rich. This data richness of India has led to over 1,000 startups working in learning, artificial intelligence and they are doing the kind of disruption that the world has never seen before. Many of them are doing path-breaking work," Kant said. He was speaking to a select audience in Washington, where he was visiting as part of his three-city US tour.

It is now well known that deep neural networks (DNNs) are vulnerable to adversarial attack. Adversarial samples are similar to the clean ones, but are able to cheat the attacked DNN to produce incorrect predictions in high confidence. But most of the existing adversarial attacks have high success rate only when the information of the attacked DNN is well-known or could be estimated by massive queries. A promising way is to generate adversarial samples with high transferability. By this way, we generate 96020 transferable adversarial samples from original ones in ImageNet. The average difference, measured by root means squared deviation, is only around 3.8 on average. However, the adversarial samples are misclassified by various models with an error rate up to 90\%. Since the images are generated independently with the attacked DNNs, this is essentially zero-query adversarial attack. We call the dataset \emph{DAmageNet}, which is the first universal adversarial dataset that beats many models trained in ImageNet. By finding the drawbacks, DAmageNet could serve as a benchmark to study and improve robustness of DNNs. DAmageNet could be downloaded in http://www.pami.sjtu.edu.cn/Show/56/122.

Detecting biases in artificial intelligence has become difficult because of the impenetrable nature of deep learning. The central difficulty is in relating unobservable phenomena deep inside models with observable, outside quantities that we can measure from inputs and outputs. For example, can we detect gendered perceptions of occupations (e.g., female librarian, male electrician) using questions to and answers from a word embedding-based system? Current techniques for detecting biases are often customized for a task, dataset, or method, affecting their generalization. In this work, we draw from Psychophysics in Experimental Psychology---meant to relate quantities from the real world (i.e., "Physics") into subjective measures in the mind (i.e., "Psyche")---to propose an intellectually coherent and generalizable framework to detect biases in AI. Specifically, we adapt the two-alternative forced choice task (2AFC) to estimate potential biases and the strength of those biases in black-box models. We successfully reproduce previously-known biased perceptions in word embeddings and sentiment analysis predictions. We discuss how concepts in experimental psychology can be naturally applied to understanding artificial mental phenomena, and how psychophysics can form a useful methodological foundation to study fairness in AI.

This is an easy prediction to make, because even Tesla isn't claiming that its eye-catching angular steel beast will be available for sale in 2020. The company's own pitch is that production won't even begin until 2021, with owners receiving their first shipments in 2022. But the gap is relevant to Tesla's future: where the company was once genuinely ahead of the curve, in making beautiful electric cars that people wanted to buy, it has increasingly relied on beating its competitors to announcements, rather than actually shipping. The list of Elon Musk's as-yet-unfulfilled promises grows every year – but the electric fleets of BMW, Ford, General Motors and others grow faster. One of the most impressive, and futuristic, products to have come from Google in recent years, Duplex is an AI assistant that can make calls to local businesses on your behalf to do things like book appointments and find out opening times.