Conventional AI evaluation approaches concentrated within the AI stack exhibit systemic limitations for exploring, navigating and resolving the human and societal factors that play out in real world deployment such as in education, finance, healthcare, and employment sectors. AI capability evaluations can capture detail about first-order effects, such as whether immediate system outputs are accurate, or contain toxic, biased or stereotypical content, but AI's second-order effects, i.e. any long-term outcomes and consequences that may result from AI use in the real world, have become a significant area of interest as the technology becomes embedded in our daily lives. These secondary effects can include shifts in user behavior, societal, cultural and economic ramifications, workforce transformations, and long-term downstream impacts that may result from a broad and growing set of risks. This position paper argues that measuring the indirect and secondary effects of AI will require expansion beyond static, single-turn approaches conducted in silico to include testing paradigms that can capture what actually materializes when people use AI technology in context. Specifically, we describe the need for data and methods that can facilitate contextual awareness and enable downstream interpretation and decision making about AI's secondary effects, and recommend requirements for a new ecosystem.

Recent studies on the safety alignment of large language models (LLMs) have revealed that existing approaches often operate superficially, leaving models vulnerable to various adversarial attacks. Despite their significance, these studies generally fail to offer actionable solutions beyond data augmentation for achieving more robust safety mechanisms. This paper identifies a fundamental cause of this superficiality: existing alignment approaches often presume that models can implicitly learn a safety-related reasoning task during the alignment process, enabling them to refuse harmful requests. However, the learned safety signals are often diluted by other competing objectives, leading models to struggle with drawing a firm safety-conscious decision boundary when confronted with adversarial attacks. Based on this observation, by explicitly introducing a safety-related binary classification task and integrating its signals with our attention and decoding strategies, we eliminate this ambiguity and allow models to respond more responsibly to malicious queries. We emphasize that, with less than 0.2x overhead cost, our approach enables LLMs to assess the safety of both the query and the previously generated tokens at each necessary generating step. Extensive experiments demonstrate that our method significantly improves the resilience of LLMs against various adversarial attacks, offering a promising pathway toward more robust generative AI systems.

Autonomous vehicles (AVs) have significantly advanced in real-world deployment in recent years, yet safety continues to be a critical barrier to widespread adoption. Traditional functional safety approaches, which primarily verify the reliability, robustness, and adequacy of AV hardware and software systems from a vehicle-centric perspective, do not sufficiently address the AV's broader interactions and behavioral impact on the surrounding traffic environment. To overcome this limitation, we propose a paradigm shift toward behavioral safety, a comprehensive approach focused on evaluating AV responses and interactions within traffic environment. To systematically assess behavioral safety, we introduce a third-party AV safety assessment framework comprising two complementary evaluation components: Driver Licensing Test and Driving Intelligence Test. The Driver Licensing Test evaluates AV's reactive behaviors under controlled scenarios, ensuring basic behavioral competency. In contrast, the Driving Intelligence Test assesses AV's interactive behaviors within naturalistic traffic conditions, quantifying the frequency of safety-critical events to deliver statistically meaningful safety metrics before large-scale deployment. We validated our proposed framework using \texttt{Autoware.Universe}, an open-source Level 4 AV, tested both in simulated environments and on the physical test track at the University of Michigan's Mcity Testing Facility. The results indicate that \texttt{Autoware.Universe} passed 6 out of 14 scenarios and exhibited a crash rate of 3.01e-3 crashes per mile, approximately 1,000 times higher than average human driver crash rate. During the tests, we also uncovered several unknown unsafe scenarios for \texttt{Autoware.Universe}. These findings underscore the necessity of behavioral safety evaluations for improving AV safety performance prior to widespread public deployment.

Ukraine staged a dramatic series of strikes across Russia, deploying drones hidden in trucks deep inside the country to hit strategic airfields as far away as eastern Siberia. Around the same time, Moscow launched one of its longest drone and missile attacks against Kyiv, escalating tensions ahead of crucial peace talks this week. More than 40 Russian aircraft, including the Tu-95 and Tu-22 M3 long-range bombers capable of deploying conventional and nuclear weapons as well as the A-50, are reported to have been damaged in the operation on Sunday, an official in Ukraine's Security Service said on condition of anonymity as the details are not public. Ukraine's Security Service chief Vasyl Malyuk led the operation and losses are assessed to be at least 2 billion, the person said.

"No intelligence operation in the world has done anything like this before," defence analyst Serhii Kuzan told Ukrainian TV. "These strategic bombers are capable of launching long-range strikes against us," he said. "There are only 120 of them and we struck 40. It is hard to assess the damage, but Ukrainian military blogger Oleksandr Kovalenko says that even if the bombers, and command and control aircraft were not destroyed, the impact is enormous. "The extent of the damage is such that the Russian military-industrial complex, in its current state, is unlikely to be able to restore them in the near future," he wrote on his Telegram channel. The strategic missile-carrying bombers in question, the Tu-95, Tu-22, and Tu-160 are, he said, no longer in production. Repairing them will be difficult, replacing them impossible. The loss of the supersonic Tu-160, he said, would be especially keenly felt. "Today, the Russian Aerospace Forces lost not just two of their rarest aircraft, but truly two unicorns in the herd," he wrote. Beyond the physical damage, which may or may not be as great as analysts here are assessing, Operation Spider's Web sends another critical message, not just to Russia but also to Ukraine's western allies. My colleague Svyatoslav Khomenko, writing for the BBC Ukrainian Service website, recalls a recent encounter with a government official in Kyiv. "The biggest problem," the official told Svyatoslav, "is that the Americans have convinced themselves we've already lost the war.

Republican South Dakota Sen. Mike Rounds spoke to Fox News Digital about the U.S.' efforts to remain the world's economic and military leader as China makes technological advances. SIMI VALLEY, CALIFORNIA – China's rampant theft of intellectual property from American institutions and industry must be thwarted as the U.S. battles to remain the world's economic and military leader – but America cannot completely decouple from the economic behemoth, Republican South Dakota Sen. Mike Rounds told Fox News Digital in an exclusive interview. The key, according to Rounds, is maintaining China as a key trade partner without giving them access to America's technology, including artificial intelligence and computer chips. "In doing so, maybe we'll restrict their ability to actually be able to have a market that they can create their own stuff with. And in using ours, they'll be our standards," Rounds told Fox News Digital in an exclusive interview from the Ronald Reagan Presidential Library during the Reagan National Economic Forum on Friday in California.

Meanwhile, Ukrainian authorities reported a massive overnight drone and missile attack on its territory. All this comes as Russian and Ukrainian negotiators are heading to Istanbul, Turkey, for a second round of peace talks on Monday. Expectations are low, as the two warring sides remain far apart on how to end the war. Russian President Vladimir Putin launched a full-scale invasion of Ukraine in February 2022. Moscow currently controls about 20% of Ukrainian territory, including the southern Crimea peninsula annexed in 2014.

Ukrainian forces destroyed dozens of Russian warplanes with a drone attack on air bases deep within Russian territory on Sunday. Ukrainian forces destroyed 40 aircraft in the attack, which an official says took more than a year to orchestrate. Russia's defense ministry confirmed the attack on Sunday, saying it struck five airfields. The operation saw drones transported in containers carried by trucks deep into Russian territory, he said. The drones reportedly hit 41 planes stationed at several airfields on Sunday afternoon, including A-50, Tu-95 and Tu-22M aircraft, the official said.

Officials say multiple military airbases have come under drone attacks in Russia in a major operation taking place ahead of peace talks with Ukraine due to start in Istanbul on Monday. The Russian Defence Ministry said that Ukraine had launched drone strikes targeting Russian military airfields across five regions on Sunday, causing several aircraft to catch fire. The attacks occurred in the Murmansk, Irkutsk, Ivanovo, Ryazan, and Amur regions. Air defences repelled the assaults in all but two regions – Murmansk and Irkutsk, the ministry said. "In the Murmansk and Irkutsk regions, the launch of FPV drones from an area in close proximity to airfields resulted in several aircraft catching fire," the ministry said.

Sen. Ted Cruz, R-Texas, joins'America's Newsroom' along with 15-year-old AI deep fake victim Elliston Berry to discuss the importance of the'Take It Down' bill, warning the issue is'rising every day.' Evangelical leaders praised President Donald Trump for his leadership on artificial intelligence ("AI") in an open letter published last week, while cautioning him to ensure the technology is developed responsibly. Dubbing Trump the "AI President," the religious leaders wrote that they believe Trump is there by "Divine Providence" to guide the world on the future of AI. The signatories said they are "pro-science" and fully support the advancement of technology which benefits their own ministries around the world. "We are also pro-economic prosperity and economic leadership for America and our friends. We do not want to see the AI revolution slowing, but we want to see the AI revolution accelerating responsibly," the letter says.