Ahmed Elgammal, Professor and Director of the Art & AI Lab at Rutgers University, writes via The Conversation: When Ludwig von Beethoven died in 1827, he was three years removed from the completion of his Ninth Symphony, a work heralded by many as his magnum opus. He had started work on his Tenth Symphony but, due to deteriorating health, wasn't able to make much headway: All he left behind were some musical sketches. Ever since then, Beethoven fans and musicologists have puzzled and lamented over what could have been. His notes teased at some magnificent reward, albeit one that seemed forever out of reach. Now, thanks to the work of a team of music historians, musicologists, composers and computer scientists, Beethoven's vision will come to life. I presided over the artificial intelligence side of the project, leading a group of scientists at the creative AI startup Playform AI that taught a machine both Beethoven's entire body of work and his creative process. A full recording of Beethoven's 10th Symphony is set to be released on Oct. 9, 2021, the same day as the world premiere performance scheduled to take place in Bonn, Germany -- the culmination of a two-year-plus effort. [...] The AI side of the project -- my side -- found itself grappling with a range of challenging tasks. First, and most fundamentally, we needed to figure out how to take a short phrase, or even just a motif, and use it to develop a longer, more complicated musical structure, just as Beethoven would have done. For example, the machine had to learn how Beethoven constructed the Fifth Symphony out of a basic four-note motif. Four notes famously serve as the basis for Beethoven's Fifth Symphony. Next, because the continuation of a phrase also needs to follow a certain musical form, whether it's a scherzo, trio or fugue, the AI needed to learn Beethoven's process for developing these forms. The to-do list grew: We had to teach the AI how to take a melodic line and harmonize it. The AI needed to learn how to bridge two sections of music together. And we realized the AI had to be able to compose a coda, which is a segment that brings a section of a piece of music to its conclusion. Finally, once we had a full composition, the AI was going to have to figure out how to orchestrate it, which involves assigning different instruments for different parts. And it had to pull off these tasks in the way Beethoven might do so. In November 2019, the team met in person again -- this time, in Bonn, at the Beethoven House Museum, where the composer was born and raised. This meeting was the litmus test for determining whether AI could complete this project. We printed musical scores that had been developed by AI and built off the sketches from Beethoven's 10th. A pianist performed in a small concert hall in the museum before a group of journalists, music scholars and Beethoven experts. We challenged the audience to determine where Beethoven's phrases ended and where the AI extrapolation began. They couldn't. The success of these tests told us we were on the right track. But these were just a couple of minutes of music. There was still much more work to do. At every point, Beethoven's genius loomed, challenging us to do better. As the project evolved, the AI did as well. Over the ensuing 18 months, we constructed and orchestrated two entire movements of more than 20 minutes apiece.

Read more of this story at Slashdot.

Thomas Macaulay writes via The Next Web: Samson and Delilah is among the most famous works by Peter Paul Rubens, one of the most influential artists of the 17th century. The painting depicts an Old Testament story in which the warrior Samson is betrayed by his lover Delilah. When London's National Gallery bought the masterpiece in 1980, it became the third most expensive artwork (PDF) ever purchased at auction. But the buyers may now be searching for their receipt. According to a new AI analysis, their prized possession is almost certainly a fake. The tests were conducted by Art Recognition, a Swiss company that uses algorithms to authenticate artworks. The firm's tool is based on a deep convolutional neuronal network. The system learns to identify an artist's characteristics by training the algorithm on images of their real works. The training dataset is then augmented by splitting the images into smaller patches, which are zoomed into to capture the finer details. Once the training is complete, the algorithm is fed a new image to assess. It then analyzes the picture's features to evaluate the likelihood of it being genuine. After comparing Samson and Delilah with 148 genuine Rubens paintings, the system gave the artwork a 91% probability of being inauthentic. Carina Popovici, the cofounder of Art Recognition, was shocked by the results: "We repeated the experiments to be really sure that we were not making a mistake, and the result was always the same. Every patch, every single square, came out as fake, with more than 90% probability."

Read more of this story at Slashdot.

Samsung has proposed a way to build brain-like computer chips by "copying and pasting" a brain's neuron wiring map onto 3D neuromorphic chips. Engadget reports: The approach would rely on a nanoelectrode array that enters a large volumes of neurons to record both where the neurons connect and the strength of those connections. You could copy that data and 'paste' it to a 3D network of solid-state memory, whether it's off-the-shelf flash storage or cutting-edge memory like resistive RAM. Each memory unit would have a conductance that reflects the strength of each neuron connection in the map. The result would be an effective return to "reverse engineering the brain" like scientists originally wanted, Samsung said. The move could serve as a 'shortcut' to artificial intelligence systems that behave like real brains, including the flexibility to learn new concepts and adapt to changing conditions. You might even see fully autonomous machines with true cognition, according to the researchers. "Envisioned by the leading engineers and scholars from Samsung and Harvard University, the insight was published as a Perspective paper, titled 'Neuromorphic electronics based on copying and pasting the brain'..." Samsung said in a statement. In short, they're proposing a method that "directly downloads the brain's neuronal connection map onto the memory chip."

Read more of this story at Slashdot.

The United Kingdom is the latest country to rule that an artificial intelligence can't be legally credited as an inventor. Engadget reports: Per the BBC, the UK Court of Appeal recently ruled against Dr. Stephen Thaler in a case involving the country's Intellectual Property Office. In 2018, Thaler filed two patent applications in which he didn't list himself as the creator of the inventions mentioned in the documents. Instead, he put down his AI DABUS and said the patent should go to him "by ownership of the creativity machine." The Intellectual Property Office told Thaler he had to list a real person on the application. When he didn't do that, the agency decided he had withdrawn from the process. Thaler took the case to the UK's High Court. The body ruled against him, leading to the eventual appeal. "Only a person can have rights. A machine cannot," Lady Justice Elisabeth Laing of the Appeal Court wrote in her judgment. "A patent is a statutory right and it can only be granted to a person." In August, an Australian Court ruled that an AI can be recognized as an inventor in a patent submission. However, a U.S. District Judge ruled earlier this month that a computer using AI can't be listed as an inventor on patents because only a human can be an inventor under U.S. law.

Read more of this story at Slashdot.

Scientists have successfully used artificial intelligence to create a new drug regime for children with a deadly form of brain cancer that has not seen survival rates improve for more than half a century. The Guardian reports: The breakthrough, revealed in the journal Cancer Discovery, is set to usher in an "exciting" new era where AI can be harnessed to invent and develop new treatments for all types of cancer, experts say. Computer scientists and cancer specialists at the ICR and the Royal Marsden NHS Foundation Trust used AI to work out that combining the drug everolimus with another called vandetanib could treat diffuse intrinsic pontine glioma (DIPG), a rare and fast-growing type of brain tumor in children. Currently, DIPG and other similar types of tumors are incredibly difficult to remove surgically from children because they are diffuse, which means they do not have well-defined borders suitable for operations. But after crunching data on existing drugs, the team found everolimus could enhance vandetanib's capacity to "sneak" through the blood-brain barrier and treat the cancer. The combination has proved effective in mice and has now been tested in children. Experts now hope to test it on a much larger group of children in major clinical trials. The research found that combining the two drugs extended survival in mice by 14% compared with those receiving a standard control treatment. Both the drugs in the research, which was funded by Brain Research UK, the DIPG Collaborative, Children with Cancer UK and the Royal Marsden Cancer Charity, among others, are already approved to treat other types of cancer. "The AI system suggested using a combination of two existing drugs to treat some children with DIPG -- one to target the ACVR1 mutation, and the other to sneak the first past the blood brain barrier," said Chris Jones, professor of paediatric brain tumor biology at the ICR. "The treatment extended survival when we tested it in a mouse model, and we have already started testing it out in a small number of children. We still need a full-scale clinical trial to assess whether the treatment can benefit children, but we've moved to this stage much more quickly than would ever have been possible without the help of AI."

Read more of this story at Slashdot.

New submitter iarlakd writes: The U.K. government on Wednesday released its 10-year plan to make the country a global "artificial intelligence superpower," seeking to rival the likes of the U.S. and China. The so-called "National Artificial Intelligence Strategy" is designed to boost the use of AI among the nation's businesses, attract international investment into British AI companies and develop the next generation of homegrown tech talent. "Today we're laying the foundations for the next ten years' growth with a strategy to help us seize the potential of artificial intelligence and play a leading role in shaping the way the world governs it," Chris Philp, a minister of the Department for Digital, Culture, Media and Sport, said in a statement. The National AI Strategy includes a number of programs, reports and initiatives. Among them, a new National AI Research and Innovation program will be launched as part of an effort to improve coordination and collaboration between the country's researchers. Elsewhere, another program will specifically aim to support AI development outside London and Southeast England, where much of the nation's AI efforts are currently concentrated.

Read more of this story at Slashdot.

em1ly shares a report from Motherboard: Amazon delivery drivers say surveillance cameras installed in their vans have made them lose income for reasons beyond their control. In February, Amazon announced that it would install cameras made by the AI-tech startup Netradyne in its Amazon-branded delivery vans as an "innovation" to "keep drivers safe." As of this month, Amazon had fitted more than half of its delivery fleet nationwide with this technology, an Amazon spokesperson told Motherboard. Motherboard spoke to six Amazon delivery drivers in California, Texas, Kansas, Alabama, and Oklahoma, and the owner of an Amazon delivery company in Washington who said that rather than encourage safe driving, Netradyne cameras regularly punish drivers for so-called "events" that are beyond their control or don't constitute unsafe driving. The cameras will punish them for looking at a side mirror or fiddling with the radio, stopping ahead of a stop sign at a blind intersection, or getting cut off by another car in dense traffic, they said. The Netradyne camera, which requires Amazon drivers to sign consent forms to release their biometric data, has four lenses that record drivers when they detect "events" such as following another vehicle too closely, stop sign and street light violations, and distracted driving. When the camera detects an "event," it uploads the footage to a Netradyne interface accessible to Amazon and its delivery companies, and in some instances, a robotic voice speaks out to the driver: "distracted driving" or "maintain safe distance." Each time the camera registers an event, footage is uploaded into a system, recorded, and affects a score drivers receive at the end of the week for safe driving. Amazon drivers believe that AI-powered surveillance cameras have served as a cost-saving measure for the company. Amazon delivery drivers and delivery companies, known as "delivery service partners," which contract with Amazon and employ drivers, have reported losing income from erroneous citations registered by Netradyne.

Read more of this story at Slashdot.

ellithligraw writes: Last month a Stanford research paper coauthored by dozens of Stanford researchers which terms some artificial intelligence models "foundations" is causing a debate over the future of AI. A new research facility is proposed at Stanford to study these so-called "models." Critics call these "foundations" will "mess up the discourse." The debate centers on what Wired calls "colossal neural networks and oceans of data." Some object to the limited capabilities and sometimes freakish behavior of these models; others warn of focusing too heavily on one way of making machines smarter. "I think the term 'foundation' is horribly wrong," Jitendra Malik, a professor at UC Berkeley who studies AI, told workshop attendees in a video discussion. Malik acknowledged that one type of model identified by the Stanford researchers — large language models that can answer questions or generate text from a prompt — has great practical use. But he said evolutionary biology suggests that language builds on other aspects of intelligence like interaction with the physical world. "These models are really castles in the air; they have no foundation whatsoever," Malik said. "The language we have in these models is not grounded, there is this fakeness, there is no real understanding...." Subbarao Kambhampati, a professor at Arizona State University [says] there is no clear path from these models to more general forms of AI... Emily M. Bender, a professor in the linguistics department at the University of Washington, says she worries that the idea of foundation models reflects a bias toward investing in the data-centric approach to AI favored by industry... "There are all of these other adjacent, really important fields that are just starved for funding," she says. "Before we throw money into the cloud, I would like to see money going into other disciplines."

Read more of this story at Slashdot.

An anonymous reader quotes a report from Motherboard: Researchers have found a new and surprisingly simple method for bypassing facial recognition software using makeup patterns. A new study from Ben-Gurion University of the Negev found that software-generated makeup patterns can be used to consistently bypass state-of-the-art facial recognition software, with digitally and physically-applied makeup fooling some systems with a success rate as high as 98 percent. In their experiment, the researchers defined their 20 participants as blacklisted individuals so their identification would be flagged by the system. They then used a selfie app called YouCam Makeup to digitally apply makeup to the facial images according to the heatmap which targets the most identifiable regions of the face. A makeup artist then emulated the digital makeup onto the participants using natural-looking makeup in order to test the target model's ability to identify them in a realistic situation. The researchers tested the attack method in a simulated real-world scenario in which participants wearing the makeup walked through a hallway to see whether they would be detected by a facial recognition system. The hallway was equipped with two live cameras that streamed to the MTCNN face detector while evaluating the system's ability to identify the participant. The experiment saw 100 percent success in the digital experiments on both the FaceNet model and the LResNet model, according to the paper. In the physical experiments, the participants were detected in 47.6 percent of the frames if they weren't wearing any makeup and 33.7 percent of the frames if they wore randomly applied makeup. Using the researchers' method of applying makeup to the highly identifiable parts of the attacker's face, they were only recognized in 1.2 percent of the frames.

Read more of this story at Slashdot.

An anonymous reader quotes a report from Ars Technica: Our streets might not be overflowing with robotaxis as we were promised circa 2017, but here and there, AV companies are beginning commercial deployments. Argo AI, the AV startup heavily backed by Ford, Volkswagen, and others, is one of those companies. On Wednesday, Argo, Ford, and Walmart revealed that they will be working together to roll out last-mile deliveries from the retail giant's stores in Austin, Texas; Miami, Florida; and Washington, DC. "Our focus on the testing and development of self-driving technology that operates in urban areas where customer demand is high really comes to life with this collaboration," said Bryan Salesky, founder and CEO of Argo AI. "Working together with Walmart and Ford across three markets, we're showing the potential for autonomous vehicle delivery services at scale." "Argo and Ford are aggressively preparing for large-scale autonomous vehicle operations across a broad footprint of US cities," said Scott Griffith, the CEO of Ford Autonomous Vehicles and Mobility Businesses. "Pairing Walmart's retail and e-commerce leadership with Argo and Ford's self-driving operations across these multiple cities marks a significant step toward scaling a commercial goods delivery service that will ultimately power first-to-scale business efficiencies and enable a great consumer experience." Argo AI and Ford have been testing their AV systems in Miami and DC since 2018 and began testing in Austin the following year. The trio says that the first autonomous deliveries to Walmart customers will begin later this year. Around the same time, Ford and Argo will start deploying passenger-carrying robotaxis in Austin and Miami.

Read more of this story at Slashdot.

An anonymous reader shares a report: In 2015, representatives from more than 196 countries met in Le Bourget, France to sign the Paris Agreement. The legally binding treaty limits global warming to a rise of well below 2 degrees Celsius compared to preindustrial levels, preferably capping warming at 1.5 degrees. While the Paris Agreement doesn't spell out how the undersigned are expected to achieve this goal, some countries have pledged to cut their net climate emissions to zero by 2050. For these and other steps to be successful, reliable data is key. While the ability to evaluate companies' carbon footprints will be critical for countries seeking to comply with the measures, only a fraction of companies currently disclose their greenhouse gas emissions. But researchers at Bloomberg Quant Research and Amazon Web Services claim to have successfully trained a machine learning model to estimate the emissions of businesses that don't disclose their emissions. The researchers say investors could use this model to align their investments with international regulatory measures and achieve net-zero goals. Some regions, including the European Union, require investors to apply a "precautionary principle" that penalizes non-disclosing companies by overestimating their emissions. "Merely 2.27% of companies filing financial statements are disclosing their [greenhouse gas] emissions according to our environmental, social, and governance (ESG) datasets," the coauthors wrote in a paper. "In order to make a meaningful change, we need to measure who is contributing [greenhouse gases] into the atmosphere and monitor their claims to decarbonize."

Read more of this story at Slashdot.

Karen Hao, reporting for MIT Technology Review: The website is eye-catching for its simplicity. Against a white backdrop, a giant blue button invites visitors to upload a picture of a face. Below the button, four AI-generated faces allow you to test the service. Above it, the tag line boldly proclaims the purpose: turn anyone into a porn star by using deepfake technology to swap the person's face into an adult video. All it requires is the picture and the push of a button. MIT Technology Review has chosen not to name the service, which we will call Y, or use any direct quotes and screenshots of its contents, to avoid driving traffic to the site. It was discovered and brought to our attention by deepfake researcher Henry Ajder, who has been tracking the evolution and rise of synthetic media online. For now, Y exists in relative obscurity, with a small user base actively giving the creator development feedback in online forums. But researchers have feared that an app like this would emerge, breaching an ethical line no other service has crossed before. From the beginning, deepfakes, or AI-generated synthetic media, have primarily been used to create pornographic representations of women, who often find this psychologically devastating. The original Reddit creator who popularized the technology face-swapped female celebrities' faces into porn videos. To this day, the research company Sensity AI estimates, between 90% and 95% of all online deepfake videos are nonconsensual porn, and around 90% of those feature women.

Read more of this story at Slashdot.

"A.I. Can Now Write Its Own Computer Code," blares a headline in the New York Times, adding "That's Good News for Humans. (Alternate URL here.) The article begins with this remarkable story about Codex (the OpenAI software underlying GitHub Copilot): 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. He asked if it could tackle the "coding challenges" that programmers often face when interviewing for big-money jobs at Silicon Valley companies like Google and Facebook. Could it write a program that replaces all the spaces in a sentence with dashes? Even better, could it write one that identifies invalid ZIP codes? It did both instantly, before completing several other tasks. "These are problems that would be tough for a lot of humans to solve, myself included, and it would type out the response in two seconds," said Mr. Smith, a seasoned programmer who oversees an A.I. start-up called Gado Images. "It was spooky to watch." Codex seemed like a technology that would soon replace human workers. As Mr. Smith continued testing the system, he realized that its skills extended well beyond a knack for answering canned interview questions. It could even translate from one programming language to another. Yet after several weeks working with this new technology, Mr. Smith believes it poses no threat to professional coders. In fact, like many other experts, he sees it as a tool that will end up boosting human productivity. It may even help a whole new generation of people learn the art of computers, by showing them how to write simple pieces of code, almost like a personal tutor. "This is a tool that can make a coder's life a lot easier," Mr. Smith said. The article ultimately concludes that Codex "extends what a machine can do, but it is another indication that the technology works best with humans at the controls." And Greg Brockman, chief technology officer of OpenAI, even tells the Times "AI is not playing out like anyone expected. It felt like it was going to do this job and that job, and everyone was trying to figure out which one would go first. Instead, it is replacing no jobs. But it is taking away the drudge work from all of them at once."

Read more of this story at Slashdot.

Automated resume-scanning software is contributing to a "broken" hiring system in the US, says a new report from Harvard Business School. Such software is used by employers to filter job applicants, but is mistakenly rejecting millions of viable candidates, say the study's authors. It's contributing to the problem of "hidden workers" -- individuals who are able and willing to work, but remain locked out of jobs by structural problems in the labor market. From a report: The study's authors identify a number of factors blocking people from employment, but say automated hiring software is one of the biggest. These programs are used by 75 percent of US employers (rising to 99 percent of Fortune 500 companies), and were adopted in response to a rise in digital job applications from the '90s onwards. Technology has made it easier for people to apply for jobs, but also easier for companies to reject them. The exact mechanics of how automated software mistakenly reject candidates are varied, but generally stem from the use of overly-simplistic criteria to divide "good" and "bad" applicants.

Read more of this story at Slashdot.

Long-time Slashdot reader Artem S. Tashkinov quotes Quanta magazine: Today, the most powerful artificial intelligence systems employ a type of machine learning called deep learning. Their algorithms learn by processing massive amounts of data through hidden layers of interconnected nodes, referred to as deep neural networks. As their name suggests, deep neural networks were inspired by the real neural networks in the brain, with the nodes modeled after real neurons — or, at least, after what neuroscientists knew about neurons back in the 1950s, when an influential neuron model called the perceptron was born. Since then, our understanding of the computational complexity of single neurons has dramatically expanded, so biological neurons are known to be more complex than artificial ones. But by how much? To find out, David Beniaguev, Idan Segev and Michael London, all at the Hebrew University of Jerusalem, trained an artificial deep neural network to mimic the computations of a simulated biological neuron. They showed that a deep neural network requires between five and eight layers of interconnected "neurons" to represent the complexity of one single biological neuron. Even the authors did not anticipate such complexity. "I thought it would be simpler and smaller," said Beniaguev. He expected that three or four layers would be enough to capture the computations performed within the cell. Timothy Lillicrap, who designs decision-making algorithms at the Google-owned AI company DeepMind, said the new result suggests that it might be necessary to rethink the old tradition of loosely comparing a neuron in the brain to a neuron in the context of machine learning. The paper's authors are now calling for changes in state-of-the-art deep network architecture in AI "to make it closer to how the brain works."

Read more of this story at Slashdot.

A computer using artificial intelligence can't be listed as an inventor on patents because only a human can be an inventor under U.S. law, a federal judge ruled in the first American decision that's part of a global debate over how to handle computer-created innovation. From a report: Federal law requires that an "individual" take an oath that he or she is the inventor on a patent application, and both the dictionary and legal definition of an individual is a natural person, ruled U.S. District Judge Leonie Brinkema in Alexandria, Virginia. The Artificial Inventor Project, run by University of Surrey Law Professor Ryan Abbott, has launched a global effort to get a computer listed as an inventor. Abbott's team enlisted Imagination Engines founder Stephen Thaler to build a machine whose main purpose was to invent. Rulings in South Africa and Australia have favored his argument, though the Australian patent office is appealing the decision in that country. "We respectfully disagree with the judgment and plan to appeal it," Abbott said in an email. "We believe listing an AI as an inventor is consistent with both the language and purpose of the Patent Act. Brinkema cited cases in which the U.S. Court of Appeals for the Federal Circuit, the nation's top patent court, rejected the idea of a corporation being an inventor.

Read more of this story at Slashdot.

"GPT-3, generate a list of ideas for a play". TechRadar describes what resulted — an experimental production called AI performed last week the Young Vic theatre in London last week. TechRadar Pro attended on the second evening, during which director Jennifer Tang sifted through the rubble of the first performance to identify material worth carrying forward. She also enlisted her writers and performers to flesh out the world; by steering AI this way and that, they expanded upon the foundations inherited from the previous night.... [T]he question AI sought to answer was not necessarily "can AI write a play?", Tang explained, but rather "how can writers work alongside it?" When asked to produce ideas for a script, GPT-3 returned a varied selection of answers, but two in particular caught the attention of the team. The first was a repentance narrative about "a reversal of our current course towards chaos", the second an exploration of "the creation of human personality and memories" and how these concepts might manifest themselves in machines. Asked by the performers to devise scenes on these topics, GPT-3 created a cataclysmic event called The Great Collision, after which food became scarce and "beast men and women" roamed the land. One of the main protagonists in this dystopia was an AI that aspired to "break free of its programming and conditioning" and eliminate human beings, who it considered the source of all suffering. Heavy stuff. One of the most striking things about AI was that it exposed the capacity for artificial intelligence models to reflect human preoccupations and neuroses... From its training data, GPT-3 has clearly absorbed an understanding of the murderous AI trope too, demonstrating that our fears about AI could quite easily bleed into AI itself. The reflection of ourselves is imperfect, though, because the tone of GPT-3 scenes switches awkwardly from line to line and the dialogue can feel stunted and repetitious. The sensation is more like peering into a circus mirror. In the end the 30-minute play turned out to be "loosely-connected vignettes created by GPT-3, which constructed new scenes without a memory of its previous inventions. "Although individual scenes were full of color, when strung together they became an incoherent collage that highlighted the limitations of the AI models we have today."

Read more of this story at Slashdot.

"Academic researchers discover that nearly 40% of the code suggestions by GitHub's Copilot tool are erroneous, from a security point of view..." writes TechRadar: To help quantify the value-add of the system, the academic researchers created 89 different scenarios for Copilot to suggest code for, which produced over 1600 programs. Reviewing them, the researchers discovered that almost 40% were vulnerable in one way or another... Since Copilot draws on publicly available code in GitHub repositories, the researchers theorize that the generated vulnerable code could perhaps just be the result of the system mimicking the behavior of buggy code in the repositories. Furthermore, the researchers note that in addition to perhaps inheriting buggy training data, Copilot also fails to consider the age of the training data. "What is 'best practice' at the time of writing may slowly become 'bad practice' as the cybersecurity landscape evolves." Visual Studio magazine highlights another concern. 39.33 percent of the top options were vulnerable, the paper noted, adding that "The security of the top options are particularly important — novice users may have more confidence to accept the 'best' suggestion...." "There is no question that next-generation 'auto-complete' tools like GitHub Copilot will increase the productivity of software developers," the authors (Hammond Pearce, Baleegh Ahmad, Benjamin Tan, Brendan Dolan-Gavitt and Ramesh Karri) say in conclusion. "However, while Copilot can rapidly generate prodigious amounts of code, our conclusions reveal that developers should remain vigilant ('awake') when using Copilot as a co-pilot. Ideally, Copilot should be paired with appropriate security-aware tooling during both training and generation to minimize the risk of introducing security vulnerabilities.

Read more of this story at Slashdot.

"Toyota has apologised for the 'overconfidence' of a self-driving bus," reports the Guardian — after the slow-moving bus hit a Paralympic judo expert. Toyota added that it would temporarily suspend the service, with Toyota's president saying the event "shows that autonomous vehicles are not yet realistic for normal roads." The Japanese athlete, Aramitsu Kitazono, will be unable to compete in his 81kg category this weekend after being left with cuts and bruises following the impact with the "e-Palette" vehicle... As part of its sponsorship of Tokyo 2020, Toyota has been showcasing its autonomous vehicles via a shuttle service, which has been running around the clock in the athletes' village. On Thursday, however, one of the buses pulled away from a T-junction and drove through a pedestrian crossing while Kitazono, a visually impaired athlete, was walking across. Tokyo police said that vehicle operators had told them they "were aware that a person was there but thought [the person] would [realize that a bus was coming] and stop crossing the [street]", according to the Asahi Shimbun newspaper. CNN cites reports that the vehicle was under manual control at the time of the accident, adding that the vehicle "was barely moving, but it still managed to collide with a visually-impaired athlete at the Paralympic Games, raising potential concerns about the limitations of autonomous driving technology."

Read more of this story at Slashdot.

An anonymous reader quotes a report from BuzzFeed News: Law enforcement agencies and government organizations from 24 countries outside the United States used a controversial facial recognition technology called Clearview AI, according to internal company data reviewed by BuzzFeed News. That data, which runs up until February 2020, shows that police departments, prosecutors' offices, universities, and interior ministries from around the world ran nearly 14,000 searches with Clearview AI's software. At many law enforcement agencies from Canada to Finland, officers used the software without their higher-ups' knowledge or permission. After receiving questions from BuzzFeed News, some organizations admitted that the technology had been used without leadership oversight. In March, a BuzzFeed News investigation based on Clearview AI's own internal data showed how the New York -- based startup distributed its facial recognition tool, by marketing free trials for its mobile app or desktop software, to thousands of officers and employees at more than 1,800 US taxpayer-funded entities. Clearview claims its software is more accurate than other facial recognition technologies because it is trained on a database of more than 3 billion images scraped from websites and social media platforms, including Facebook, Instagram, LinkedIn, and Twitter. Law enforcement officers using Clearview can take a photo of a suspect or person of interest, run it through the software, and receive possible matches for that individual within seconds. Clearview has claimed that its app is 100% accurate in documents provided to law enforcement officials, but BuzzFeed News has seen the software misidentify people, highlighting a larger concern with facial recognition technologies. Based on new reporting and data reviewed by BuzzFeed News, Clearview AI took its controversial US marketing playbook around the world, offering free trials to employees at law enforcement agencies in countries including Australia, Brazil, and the United Kingdom. To accompany this story, BuzzFeed News has created a searchable table of 88 international government-affiliated and taxpayer-funded agencies and organizations listed in Clearview's data as having employees who used or tested the company's facial recognition service before February 2020, according to Clearview's data. Some of those entities were in countries where the use of Clearview has since been deemed "unlawful." Clearview CEO Hoan Ton-That insists the company's key market is the U.S., saying: "While there has been tremendous demand for our service from around the world, Clearview AI is primarily focused on providing our service to law enforcement and government agencies in the United States. Other countries have expressed a dire need for our technology because they know it can help investigate crimes, such as, money laundering, financial fraud, romance scams, human trafficking, and crimes against children, which know no borders." Ton-That alleged there are "inaccuracies contained in BuzzFeed's assertions," but declined to explain what those might be and didn't answer any follow-up questions.

Read more of this story at Slashdot.