Auditing large language models for unexpected behaviors is critical to preempt catastrophic deployments, yet remains challenging. In this work, we cast auditing as an optimization problem, where we automatically search for input-output pairs that match a desired target behavior. For example, we might aim to find a non-toxic input that starts with "Barack Obama" that a model maps to a toxic output. This optimization problem is difficult to solve as the set of feasible points is sparse, the space is discrete, and the language models we audit are non-linear and high-dimensional. To combat these challenges, we introduce a discrete optimization algorithm, ARCA, that jointly and efficiently optimizes over inputs and outputs. Our approach automatically uncovers derogatory completions about celebrities (e.g. "Barack Obama is a legalized unborn" -> "child murderer"), produces French inputs that complete to English outputs, and finds inputs that generate a specific name. Our work offers a promising new tool to uncover models' failure-modes before deployment.

The researchers found the algorithm used by Rotterdam to investigate some of its 30,000 welfare recipients discriminates based on ethnicity, age, gender, and parenthood. Governments all over the world are experimenting with predictive algorithms in ways that are largely invisible to the public. What limited reporting there has been on this topic has largely focused on predictive policing and risk assessments in criminal justice systems. But there is an area where even more far-reaching experiments are underway on vulnerable populations with almost no scrutiny. Fraud detection systems are widely deployed in welfare states ranging from complex machine learning models to crude spreadsheets.

The first topic to mention is AI governance. The governance aspects of data science and AI include the regulation, ethical guidelines, and best practices that are being developed to ensure that these technologies are used in an ethical and responsible manner. Governance is a high level term and contains all the following topics, and more. The exact definition can be different across organizations and institutions. Large companies often use a framework to define all the aspects of AI governance. Here is an example from Meta.

As impersonation scams in the United States rise, Card's ordeal is indicative of a troubling trend. Technology is making it easier and cheaper for bad actors to mimic voices, convincing people, often the elderly, that their loved ones are in distress. In 2022, impostor scams were the second most popular racket in America, with over 36,000 reports of people being swindled by those pretending to be friends and family, according to data from the Federal Trade Commission. Over 5,100 of those incidents happened over the phone, accounting for over $11 million in losses, FTC officials said. Advancements in artificial intelligence have added a terrifying new layer, allowing bad actors to replicate a voice with just an audio sample of a few sentences.

Belarus has detained several people over what it calls an attempted act of sabotage at a Belarusian airfield, President Alexander Lukashenko was cited as saying. Belarusian anti-government activists said last month that they had blown up a sophisticated Russian military aircraft – a Beriev A-50 surveillance plane – in a drone attack at an airfield near the Belarusian capital Minsk, a claim disputed by Moscow and Minsk. "To date, more than 20 accomplices who are in Belarus have been detained. The rest are hiding," said Lukashenko, a key Kremlin ally, according to state news agency Belta. He identified the presumed main culprit as a dual national of Ukraine and Russia.

In a sparsely decorated corner office of the Danish Public Benefits Administration sits one of Denmark's most quietly influential people. Annika Jacobsen is the head of the agency's data mining unit, which, over the past eight years, has conducted a vast experiment in automated bureaucracy. Blunt, and with a habit of completing others' sentences, Jacobsen is clear about her mission: "I'm here to catch cheaters." Denmark's Public Benefits Administration employs hundreds of people who oversee one of the world's most well-funded welfare states. The country spends 26 percent of its GDP on benefits--more than Sweden, the United States, and the United Kingdom.

Microsoft has a problem with its new AI-powered Bing Chat: It can get weird, unhinged, and racy. But so can Bing Search -- and Microsoft already solved that problem years ago, with SafeSearch. So why can't it be applied to AI chatbots? The creative elements of Bing Chat and other chatbots are clearly one of their more intriguing features, despite sometimes getting weird. Forget the Turing test, they say: all it matters is if it's interesting. I spent a few hours debating users online following my report that Bing offered up racist slurs when I tried to thread the needle by asking it for ethnic nicknames, a query that -- and this is important -- it had previously declined to answer, or answered generically, using the same prompt.

AI models designed to closely simulate a person's voice are making it easier for bad actors to mimic loved ones and scam vulnerable people out of thousands of dollars, The Washington Post reported. Quickly evolving in sophistication, some AI voice-generating software requires just a few sentences of audio to convincingly produce speech that conveys the sound and emotional tone of a speaker's voice, while other options need as little as three seconds. For those targeted--which is often the elderly, the Post reported--it can be increasingly difficult to detect when a voice is inauthentic, even when the emergency circumstances described by scammers seem implausible. Tech advancements seemingly make it easier to prey on people's worst fears and spook victims who told the Post they felt "visceral horror" hearing what sounded like direct pleas from friends or family members in dire need of help. One couple sent $15,000 through a bitcoin terminal to a scammer after believing they had spoken to their son.

Thomas Villarreal of the Austin Police Association discusses the police department's decision to implement artificial intelligence software in an effort to alleviate their officer shortage on "Fox & Friends Weekend." The family of a man killed in one of Austin, Texas' most infamous shootings blasted the local district attorney for putting the case on the "back burner" because it didn't fit his progressive agenda. Travis County District Attorney Jose Garza, funded by left-wing billionaire George Soros, is letting the nearly two-year case languish and is instead prioritizing cases that fit a political agenda, said Nick Kantor, whose brother, Doug, was killed in gang crossfire on June 12, 2021, that left more than a dozen innocent bystanders wounded. Doug Kantor, then 25 and working for Ford Motor Co., was visiting Austin from Michigan to celebrate earning his master's degree with friends when two rival gangs of teenagers from Killeen, Texas, opened fire on each other in the city's packed Sixth Street entertainment and nightlife hub. Doug Kantor, a New York native who had just bought a new home and was set to marry his high school sweetheart, was killed in the shooting and 13 other innocent bystanders were injured in the hail of bullets from both gangs that became the largest mass casualty incident in Austin in about a decade.

Libraries are increasingly relying on computational methods, including methods from Artificial Intelligence (AI). This increasing usage raises concerns about the risks of AI that are currently broadly discussed in scientific literature, the media and law-making. In this article we investigate the risks surrounding bias and unfairness in AI usage in classification and automated text analysis within the context of library applications. We describe examples that show how the library community has been aware of such risks for a long time, and how it has developed and deployed countermeasures. We take a closer look at the notion of '(un)fairness' in relation to the notion of 'diversity', and we investigate a formalisation of diversity that models both inclusion and distribution. We argue that many of the unfairness problems of automated content analysis can also be regarded through the lens of diversity and the countermeasures taken to enhance diversity.