The US National Highway Traffic Safety Administration (NHTSA) has initiated an investigation of Tesla's Autopilot system. The probe follows 11 crashes with parked first responder vehicles since 2018, which resulted in 17 injuries and one death. "Most incidents took place after dark and the crash scenes encountered included scene control measures such as first responder vehicle lights, flares, an illuminated arrow board, and road cones," the NHTSA's Office of Defects Investigation (ODI) wrote in a document detailing the investigation. "The involved subject vehicles were all confirmed to have been engaged in either Autopilot or Traffic Aware Cruise Control during the approach to the crashes." That covers around 765,000 Tesla electric vehicles, as Bloomberg notes.

Artificial intelligence makes decisions using complex neural networks, genetic algorithms and other techniques. While these strategies tend to produce better results, their sheer complexity makes it difficult to understand what's happening under the hood. The "black box" algorithms may include inherent biases or be unprepared to cope with "black swan" events. Artificial intelligence also requires an extensive use of technology in general, which can increase a firm's risk exposure. One example would be a cybersecurity breach that exposes sensitive data, but other examples might include data loss that impacts the efficacy of AI algorithms or the loss of an algorithm to a competitor that hurts a firm's competitive edge.

The US government has opened a formal investigation into Tesla's Autopilot partially automated driving system after a series of collisions with parked emergency vehicles. The investigation covers 765,000 vehicles, almost everything that Tesla has sold in the US since the start of the 2014 model year. Of the crashes identified by the National Highway Traffic Safety Administration (NHTSA) as part of the investigation, 17 people were injured and one was killed. NHTSA says it has identified 11 crashes since 2018 in which Teslas on Autopilot or Traffic Aware Cruise Control have hit vehicles at scenes where first responders used flashing lights, flares, an illuminated arrow board or cones warning of hazards. The agency announced the action on Monday in a posting on its website.

U.S. auto-safety regulators have launched an investigation into Tesla's partially self-driving car system after nearly a dozen reports of the company's vehicles crashing into cars at the scenes of incidents involving emergency responders. The National Highway Traffic Safety Administration on Friday opened the probe into Tesla's Autopilot, which automatically steers, brakes and accelerates the vehicle on most roads with lanes. While the system can drive the vehicle on its own in many circumstances, drivers are supposed to keep their hands on the wheel in case they need to take over when Autopilot encounters a situation that's too complex for it to handle on its own. But Autopilot has come under scrutiny on a number of occasions in recent years. The National Transportation Safety Board and NHTSA have investigated Autopilot multiple times, including for a 2016 crash that killed a man in Florida who authorities said had too much confidence in the system's capabilities.

Researchers in the US have developed an adversarial attack against the ability of machine learning systems to correctly interpret what they see – including mission-critical items such as road signs – by shining patterned light onto real world objects. In one experiment, the approach succeeded in causing the meaning of a'STOP' roadside sign to be transformed into a '30mph' speed limit sign. Perturbations on a sign, created by shining crafted light on it, distorts how it is interpreted in a machine learning system. The research is entitled Optical Adversarial Attack, and comes from Purdue University in Indiana. An OPtical ADversarial attack (OPAD), as proposed by the paper, uses structured illumination to alter the appearance of target objects, and requires only a commodity projector, a camera and a computer. The researchers were able to successfully undertake both white-box and black box attacks using this technique.

Banks, consumer advocates and think tanks are weighing in to federal bank regulators about potential pitfalls in the use of artificial intelligence and machine learning in making loan decisions. In responses to regulators' call for comments, many expressed interest in an increased use of AI and machine learning in the banking business, along with caveats about fair lending and unlawful discrimination concerns. FinRegLab, a Washington-based research group that says it has launched a broad inquiry into the use of AI in financial services, told the agencies that machine learning could be "transformational," as current gaps "increase the cost or risk of serving particular consumer and small-business populations using traditional models and data." At the same time, the predictive power of machine learning models can increase potential risks "due to the models' greater complexity and to their potential to exacerbate historical disparities and flaws in underlying data," FinRegLab said. AI and machine learning might amplify patterns of historical discrimination and financial exclusion through reliance on flawed data or mistakes in development.

When working with their clients Accenture under Tricarico's guidance focuses on "on guiding (their) clients to more safely scale their use of AI, and build a culture of confidence within their organizations." Not all companies have an established north star of AI use. Companies and partners like Accenture are vital to these companies and their proper and ethical use of the technology.

This document provides a brief description of the National Institute of Standards and Technology (NIST) speaker recognition evaluation (SRE) conversational telephone speech (CTS) Superset. The CTS Superset has been created in an attempt to provide the research community with a large-scale dataset along with uniform metadata that can be used to effectively train and develop telephony (narrowband) speaker recognition systems. It contains a large number of telephony speech segments from more than 6800 speakers with speech durations distributed uniformly in the [10s, 60s] range. The segments have been extracted from the source corpora used to compile prior SRE datasets (SRE1996-2012), including the Greybeard corpus as well as the Switchboard and Mixer series collected by the Linguistic Data Consortium (LDC). In addition to the brief description, we also report speaker recognition results on the NIST 2020 CTS Speaker Recognition Challenge, obtained using a system trained with the CTS Superset. The results will serve as a reference baseline for the challenge.

This paper aims to explain adversarial attacks in terms of how adversarial perturbations contribute to the attacking task. We estimate attributions of different image regions to the decrease of the attacking cost based on the Shapley value. We define and quantify interactions among adversarial perturbation pixels, and decompose the entire perturbation map into relatively independent perturbation components. The decomposition of the perturbation map shows that adversarially-trained DNNs have more perturbation components in the foreground than normally-trained DNNs. Moreover, compared to the normally-trained DNN, the adversarially-trained DNN have more components which mainly decrease the score of the true category. Above analyses provide new insights into the understanding of adversarial attacks.

Prisoners across the US could soon be subjected to a high-risk application of automated surveillance. A congressional committee is pressing the Department of Justice to explore the federal use of AI to analyze inmate's phone calls. The panel has called for further research into the tech's potential to prevent suicide and violent crime, Reuters reports. Attend the tech festival of the year and get your super early bird ticket now! The system transcribes phone conversations, analyzes the tone of voice, and detects certain words or phrases that are pre-programmed by officials.