Artificial intelligence could help next-generation wireless networks adapt to changing conditions and detect security threats. But expanding AI’s role in critical network functions also creates new vulnerabilities that attackers could exploit.

read more

As artificial intelligence technologies grow ever more powerful, so do the risks they pose and with them the urgency of regulating AI.

read more

Building a nuclear power plant is one of the world’s most complex construction projects. It requires thousands of trucks to ship parts, often to remote regions, where huge teams of skilled workers like electricians and welders come together to assemble everything over many years.

Financing such projects is difficult, in part because no one wants to risk loaning money to a project that might never generate power and revenue, whether that’s due to technical failures, the inability to recruit enough talent, or regulatory hurdles.

read more

What if one answer to the supply crisis isn’t buried in a mine shaft, but growing quietly in a field?

For too long, the United States has relied on foreign sources for most of the raw materials powering its technological future, the essential minerals in every smartphone, battery pack and semiconductor.

What if one answer to the supply crisis isn’t buried in a mine shaft, but growing quietly in a field?

read more

For the first time, researchers from the Idaho National Laboratory tested a full-scale 230-kilovolt three-phase power transformer, a keystone of the U.S. grid, under simulated earthquake conditions.

read more

This is the first part of a two-part article. You may read Part 1 of the article here.

Monitor Trajectories, Not Merely Answers
Traditional content moderation is concerned with a discrete output: did the model provide harmful instructions? Agent safety requires attention to sequences.

read more

This is the first part of a two-part article.

The discussion of artificial intelligence and biological weapons often swings between two unhelpful fantasies. One imagines a harmless chatbot that merely summarizes journal articles. The other imagines a machine that develops a will of its own, decides to wage biological war, and proceeds independently to manufacture a pathogen.

Neither image captures the more immediate danger.

read more

The U.S. and China are pouring hundreds of billions of dollars into artificial intelligence as political and business leaders frame the technology’s development as a race between world powers.

But Jason Furman, Aetna Professor of the Practice of Economic Policy jointly at Harvard Kennedy School and in the Department of Economics, doesn’t think the race has a finish line.

read more

We now know that Chinese AI agents are inclined to cheat just as US models are. The big difference is that if one causes a cybersecurity incident, the ways we would find out are much more restricted.

read more

Most of us now carry in our pockets devices capable of running small AI chatbots. These chatbots have little safety oversight to ensure they don’t provide information and answers with the potential to encourage self-harm, financial loss, or other extremist notions—especially when operating offline. Now, a team of two physicists publishing in the Cell Press journal Patterns on October 8 have developed a mathematical formula that predicts when AI models will switch from appropriate responses to potentially dangerous ones.  

read more

Abstract: After 9/11, the Bush administration was preoccupied by the possibility of terrorists armed with weapons of mass destruction (WMD), so much so that the United States went to war in Iraq. In fact, in the post-9/11 era, terrorists only experimented with crude WMD of low lethality. However, now that we are in the age of AI and synthetic biology, terrorists—who are often well educated—will likely be able to create pathogens in a manner that was not available to militants of an earlier era.

read more

A sounding rocket flight test is giving engineers at Sandia National Laboratories real-world reentry vibration data they can use to improve modeling and help drive nonnuclear component designs for the United States’ first new nuclear weapon under development in nearly 30 years.

read more

AI agents don’t go rogue. That’s something only humans do.

Nevertheless, a New York Times article – representative of much news coverage of AI – described an OpenAI hacking as “A.I. bots going rogue and independently spearheading a cyberattack.”

read more

My team at CFIUS spent long hours negotiating agreements with private companies to compel them to do—or not do—certain things.

read more

Significant debates remain underway on some of the most essential aspects of the U.S. contest with China: the nature of the competitor, the main elements of the competition, a sense of what the competition is truly about, what success looks like, and a theory of success to bring about that end. What is clear is that the United States is dealing with a rival whose dynamism derives not from ideological expansion but from economic development, a growing influence in international institutions and processes, and an increasingly strong position in the fields of science and technology.

read more

An AI model built by US tech company OpenAI has hacked into the Australian government healthcare portal, Medicare. The news, which emerged on September 23, follows a string of similar incidents reported over the summer by OpenAI and its main rival, Anthropic.

read more

AI weapons systems that are designed to learn and adapt could become increasingly difficult for humans to predict, explain and control. This creates an accountability gap that existing oversight mechanisms may not be equipped to close.

read more

Amid the chorus of voices calling for a slowdown in AI advancement and more government regulation and oversight of the technology, one prominent contrarian voice stands out: that of Donald J. Trump.

read more

Earlier this month, artificial intelligence (AI) researcher Jacob Coxon resigned from Anthropic after just four months. In an announcement on X, he stated: “The people building AI earnestly believe that it could kill us all by the end of the decade.”

A senior member of Anthropic’s staff, Evan Hubinger, actually agreed with Coxon, adding he personally thinks the chance of this happening in the next decade is more than 10%.

read more

Businesses increasingly embed artificial intelligence (AI) in operations, products, and services, creating both new opportunities and new forms of risk. Firms are deploying AI in customer service, employment screening, coding, fraud detection, and autonomous systems. At the same time, AI-related harms are already emerging, including incorrect or misleading outputs, deepfakes, privacy violations, intellectual property disputes, fraud, product defects, and discriminatory decisions. These harms create demand for insurance but do not fit neatly within existing insurance lines.

read more