Former El Paso U.S. Marshal Robert Almonte reacts to the latest report on border encounters from CBP. FIRST ON FOX: Sen. Chuck Grassley, R-Iowa., is seeking information from top border and homeland security agencies about the potential threat posed by drones operated by terrorist groups and cartels at the southern border amid heightened awareness of a terror threat in recent weeks. Grassley sent letters to Customs and Border Protection (CBP), Immigration and Customs Enforcement's (ICE) Homeland Security Investigations (HSI) and the Bureau of Alcohol, Tobacco, Firearms and Explosives (ATF) citing reports that Mexican cartels have increased their use of the drones at both sides of the U.S.-Mexico border. "These drones are used by the cartels to carry out targeted assassinations and violent attacks by dropping explosives in Mexico, monitor and gain reconnaissance on the movements of U.S. Border Patrol agents and other U.S. law enforcement officers, and track the progress of their smugglers illegally crossing into the U.S.," he said. Senator Chuck Grassley, R-Iowa, wrote to Customs and Border Protection, Immigration and Customs Enforcement's Homeland Security Investigations and the Bureau of Alcohol, Tobacco, Firearms and Explosives. Department of Homeland Security has noted the use of drones by cartels as a threat to the U.S. in its FY 24 threat assessment.

The UK's AI Safety Summit closes today, with UK prime minister Rishi Sunak convening a small group of world and business leaders to discuss the risks of AI. But perhaps the most surprising development is a conversation due to take place this evening between Sunak and entrepreneur Elon Musk. Full details of the event are still unclear, but it will be broadcast on X, Musk's social media platform, on a delay. The conversation is scheduled to last 45 minutes and will reportedly be moderated by an unknown individual, with some audience members, including tech executives and journalists, invited to ask questions. Politico reports that the UK government has said that the conversation won't be edited before broadcast.

This week, UK prime minister Rishi Sunak is hosting a group of more than 100 representatives from the worlds of business and politics to discuss the potential and pitfalls of artificial intelligence. The AI Safety Summit, held at Bletchley Park, UK, began on 1 November and aims to come up with a set of global principles with which to develop and deploy "frontier AI models" – the terminology favoured by Sunak and key figures in the AI industry for powerful models that don't yet exist, but may be built very soon. While the Bletchley Park event is the focal point, there is a wider week of fringe events being held in the UK, alongside a raft of UK government announcements on AI. Here are the latest developments. The key outcome of the first day of the AI Safety Summit yesterday was the Bletchley Declaration, which saw 27 countries and the European Union agree to meet more in the future to discuss the risks of AI. The UK government was keen to tout the agreement as a massive success, while impartial observers were more muted about the scale of its achievement.

A map of the country is on the screen, overlaid with a honeycomb pattern of hexagonal tiles, ranging from pale yellow to blood red. As the group types questions into a chatbot, filtering for areas close to schools or power lines, the model zooms into the satellite imagery until a field with individual trees becomes visible. A red bubble with an exclamation point marks a suspected landmine. A staffer clicks a button, creating a request to dispatch a demining team to clear it. More than 600 days since Russia's invasion, Ukraine has surpassed Afghanistan and Syria to become the most heavily mined country on earth.

In a bustling restaurant in downtown Anytown, USA, an overwhelmed manager turns to AI to help with staff shortages and customer service. Across town, a harried newspaper publisher leverages AI to help generate news content. Both are part of a growing number who rely on AI for everyday business needs. But what happens when the technology errs, or worse, poses risks we haven't fully considered? The current policy conversation is heavily geared toward the eight or so powerful companies that make AI.

Training artificial intelligence models does not typically involve coming face-to-face with an armed soldier who is pointing a gun at you and shouting at your driver to get out of the car. But the system that F. LeRon Shults and Justin Lane, cofounders of CulturePulse, are developing for the United Nations is not a typical AI model. "I got pulled over by the [Israeli] military, by a guy holding [a military rifle] because we had a Palestinian taxi driver who drove past a line he wasn't supposed to," Shults tells WIRED. "So that was an adventure." Shults and Lane were in the West Bank in September, just weeks before Hamas attacked Israel on October 7, sparking what has become one of the worst periods of violence in the region in at least 50 years.

Ukraine's Interior Minister Ihor Klymenko said 118 settlements in 10 regions of Ukraine's east had come under Russian fire in the previous 24 hours, marking the heaviest day of Russian shelling this year. Ukraine said the Kremenchuk oil refinery in central Ukraine caught fire after a Russian drone attack that knocked out the power supply in three villages while falling debris from downed drones damaged railway power lines in a nearby region. Officials said the fire was quickly extinguished. Ukraine's air force said air defences shot down 18 of 20 Russian drones and a missile before they reached their targets. Writing in The Economist newspaper, Ukraine's commander-in-chief General Valery Zaluzhny said the army needed new military capabilities and technological innovation – and air power, in particular – to break out of the current attritional fighting along the front line.

China has agreed to work with the United States, European Union and other countries to collectively manage the risk from artificial intelligence at a British summit on Wednesday aimed at charting a safe way forward for the rapidly evolving technology. Some tech executives and political leaders have warned that the rapid development of AI poses an existential threat to the world if not controlled, sparking a race by governments and international institutions to design safeguards and regulations. In a first for Western efforts to manage its safe development, a Chinese vice minister joined U.S. and EU leaders and tech bosses such as Elon Musk and ChatGPT's Sam Altman at Bletchley Park, home of Britain's World War Two code-breakers.

Sampling a probability distribution with an unknown normalization constant is a fundamental problem in computational science and engineering. This task may be cast as an optimization problem over all probability measures, and an initial distribution can be evolved to the desired minimizer dynamically via gradient flows. Mean-field models, whose law is governed by the gradient flow in the space of probability measures, may also be identified; particle approximations of these mean-field models form the basis of algorithms. The gradient flow approach is also the basis of algorithms for variational inference, in which the optimization is performed over a parameterized family of probability distributions such as Gaussians, and the underlying gradient flow is restricted to the parameterized family. By choosing different energy functionals and metrics for the gradient flow, different algorithms with different convergence properties arise. In this paper, we concentrate on the Kullback-Leibler divergence after showing that, up to scaling, it has the unique property that the gradient flows resulting from this choice of energy do not depend on the normalization constant. For the metrics, we focus on variants of the Fisher-Rao, Wasserstein, and Stein metrics; we introduce the affine invariance property for gradient flows, and their corresponding mean-field models, determine whether a given metric leads to affine invariance, and modify it to make it affine invariant if it does not. We study the resulting gradient flows in both probability density space and Gaussian space. The flow in the Gaussian space may be understood as a Gaussian approximation of the flow. We demonstrate that the Gaussian approximation based on the metric and through moment closure coincide, establish connections between them, and study their long-time convergence properties showing the advantages of affine invariance.

While Large Language Models (LLMs) are increasingly being used in real-world applications, they remain vulnerable to prompt injection attacks: malicious third party prompts that subvert the intent of the system designer. To help researchers study this problem, we present a dataset of over 126,000 prompt injection attacks and 46,000 prompt-based "defenses" against prompt injection, all created by players of an online game called Tensor Trust. To the best of our knowledge, this is currently the largest dataset of human-generated adversarial examples for instruction-following LLMs. The attacks in our dataset have a lot of easily interpretable stucture, and shed light on the weaknesses of LLMs. We also use the dataset to create a benchmark for resistance to two types of prompt injection, which we refer to as prompt extraction and prompt hijacking. Our benchmark results show that many models are vulnerable to the attack strategies in the Tensor Trust dataset. Furthermore, we show that some attack strategies from the dataset generalize to deployed LLM-based applications, even though they have a very different set of constraints to the game. We release all data and source code at https://tensortrust.ai/paper