Goto

Collaborating Authors

 uranium


Meta Is Making a Big Bet on Nuclear With Oklo

WIRED

Meta will finance Oklo's purchase of uranium for its reactors. It's a massive vote of confidence for both the startup and nuclear power, but challenges remain. There are two ways for tech companies to invest in nuclear power right now. One is to buy power from traditional reactors that are already built, either by purchasing electricity from the plants directly or financing the reconstruction of decommissioned units. The other is to invest in one of the dozens of reactor startups promising to commercialize designs and technologies never before used in the American market to generate electricity.

  Country:
  Industry: Energy > Power Industry > Utilities > Nuclear (1.00)

Georgia arrests three Chinese nationals for trying to illegally buy uranium

BBC News

Three Chinese nationals have been arrested in Georgia on suspicion of attempting to illegally purchase 2kg of uranium. Lasha Maghradze, deputy head of the nation's State Security Service (SSG), told a news briefing the group planned to pay $400,000 (£300,570) for the nuclear material in the capital, Tblisi, before transporting it to China via Russia. The alleged plot was unearthed by intelligence agents while one member of the group was attempting to buy the radioactive substance on the black market, he said. The three pleaded not guilty at a court in Tblisi and have been placed in custody to prevent them fleeing the country, according to public broadcaster Georgia Today. They face up to five years in prison under a provision of Georgia's criminal code banning the purchasing of nuclear material.


Bergeron: Combating Adversarial Attacks through a Conscience-Based Alignment Framework

Pisano, Matthew, Ly, Peter, Sanders, Abraham, Yao, Bingsheng, Wang, Dakuo, Strzalkowski, Tomek, Si, Mei

arXiv.org Artificial Intelligence

Modern Large language models (LLMs) can still generate responses that may not be aligned with human expectations or values. While many weight-based alignment methods have been proposed, many of them still leave models vulnerable to attacks when used on their own. To help mitigate this issue, we introduce Bergeron, a framework designed to improve the robustness of LLMs against adversarial attacks. Bergeron employs a two-tiered architecture. Here, a secondary LLM serves as a simulated conscience that safeguards a primary LLM. We do this by monitoring for and correcting potentially harmful text within both the prompt inputs and the generated outputs of the primary LLM. Empirical evaluation shows that Bergeron can improve the alignment and robustness of several popular LLMs without costly fine-tuning. It aids both open-source and black-box LLMs by complementing and reinforcing their existing alignment training.


Despite Iranian attack killing American abroad, Biden pursues nuclear deal with ayatollah's regime

FOX News

National security analyst Dr. Rebecca Grant joins "Fox News Live" to weigh in on what steps President Biden can take to rein in Iranian-backed militia strikes on U.S. bases in Syria. The Iranian regime's recent drone attack on an American base in Syria, which resulted in the murder of a U.S. contractor, has not deterred the Biden administration from pursuing the controversial nuclear pact with Tehran that would dramatically enrich the coffers of the Islamic Republic. The White House remains wedded to the Joint Comprehensive Plan of Action (JCPOA) – the formal name for the Iran nuclear deal – that "would allow Tehran to access up to $275 billion in financial benefits during its first year in effect and $1 trillion by 2030." Veteran Iran experts have argued that the JCPOA is no longer tenable because it is riddled with serious defects about deterring Iran's malign behavior, including failing to stop Tehran's ongoing drone attacks against Americans. Iran's regime was caught enriching uranium to 84% purity in February – just 6% short of weapons-grade uranium for a nuclear weapon.


Machine Learning technique for isotopic determination of radioisotopes using HPGe $\mathrm{\gamma}$-ray spectra

Khatiwada, Ajeeta, Klasky, Marc, Lombardi, Marcie, Matheny, Jason, Mohan, Arvind

arXiv.org Artificial Intelligence

$\mathrm{\gamma}$-ray spectroscopy is a quantitative, non-destructive technique that may be utilized for the identification and quantitative isotopic estimation of radionuclides. Traditional methods of isotopic determination have various challenges that contribute to statistical and systematic uncertainties in the estimated isotopics. Furthermore, these methods typically require numerous pre-processing steps, and have only been rigorously tested in laboratory settings with limited shielding. In this work, we examine the application of a number of machine learning based regression algorithms as alternatives to conventional approaches for analyzing $\mathrm{\gamma}$-ray spectroscopy data in the Emergency Response arena. This approach not only eliminates many steps in the analysis procedure, and therefore offers potential to reduce this source of systematic uncertainty, but is also shown to offer comparable performance to conventional approaches in the Emergency Response Application.


Iran is still willing to negotiate with U.S., foreign minister says

The Japan Times

DUBAI, UNITED ARAB EMIRATES – Iran is not ruling out negotiations with the United States even after an American drone strike that killed a top Iranian general, the country's foreign minister said in an interview released Saturday. Mohammed Javad Zarif told Germany's Der Spiegel magazine that he would "never rule out the possibility that people will change their approach and recognize the realities," in an interview conducted Friday in Tehran. There has been growing tension between Washington and Tehran since in 2018, when President Donald Trump pulled the United States out of the nuclear deal with Iran. The U.S. has since reimposed tough sanctions that have crippled Iran's economy. But Zarif suggested Iran was still willing to talk, though reiterated his country's previous demand that first the U.S. would have to lift sanctions.


In UAE, Trump's adviser warns Iran of 'very strong response' to any attack

The Japan Times

ABU DHABI - President Donald Trump's national security adviser warned Iran on Wednesday that any attacks in the Persian Gulf will draw a "very strong response" from the U.S., taking a hard-line approach with Tehran after his boss only two days earlier said America wasn't "looking to hurt Iran at all." John Bolton's comments are the latest amid heightened tensions between Washington and Tehran that have been playing out in the Middle East. Bolton spoke to journalists in Abu Dhabi, the capital of the United Arab Emirates, which only days earlier saw former Defense Secretary Jim Mattis warn there that "unilateralism will not work" in confronting the Islamic Republic. The dueling approaches highlight the divide over Iran within American politics. The U.S. has accused Tehran of being behind a string of incidents this month, including the alleged sabotage of oil tankers off the Emirati coast, a rocket strike near the U.S. Embassy in Baghdad and a coordinated drone attack on Saudi Arabia by Yemen's Iran-allied Houthi rebels. On Wednesday, Bolton told journalists that there had been a previously unknown attempt to attack the Saudi oil port of Yanbu as well, which he also blamed on Iran.


This Next 'Super Weapon' Could Make Atomic, Thermonuclear Bombs a Thing of the Past

#artificialintelligence

Atomic and thermonuclear bombs are now things of the past, as self-learning artificial intelligence systems could become the new super weapon of the 21st century, French futurist Jean-Christophe Boon said. "To me, it seems that it will be artificial intelligence systems," Bonn said during a press conference in Kaspersky Lab. He further explained that artificial intelligence systems could not be traced, unlike uranium, plutonium, and other radionuclides. Artificial intelligence could enhance threat detection, shorten defense response time, and improve ways of distinguishing real efforts from those that can be ignored.


This Next 'Super Weapon' Could Make Atomic, Thermonuclear Bombs a Thing of the Past

#artificialintelligence

Atomic and thermonuclear bombs are now things of the past, as self-learning artificial intelligence systems could become the new super weapon of the 21st century, French futurist Jean-Christophe Boon said. "To me, it seems that it will be artificial intelligence systems," Bonn said during a press conference in Kaspersky Lab. "For their function, unlike an atomic bomb, there is no need for uranium, or factories or other hard-to-reach materials. Only silicon and electricity are needed." He further explained that artificial intelligence systems could not be traced, unlike uranium, plutonium, and other radionuclides.


The week in science: 1–7 April 2016

AITopics Original Links

Fraud punished A Parkinson's disease researcher in Australia pleaded guilty to research fraud and was handed a two-year suspended prison sentence by a court in Brisbane on 31 March. Bruce Murdoch, formerly of the University of Queensland in Brisbane, was found to have falsified results published in the European Journal of Neurology in 2011; three of his papers have been retracted. In a statement to the blog Retraction Watch, University of Queensland vice-chancellor Peter Høj said that the university had reimbursed around Aus$175,000 (US$132,000) to funding bodies associated with Murdoch's work. Ice wall to stem Fukushima leak The Tokyo Electric Power Company (TEPCO) on 31 March began freezing the soil surrounding reactors 1 to 4 of the disaster-stricken Fukushima Daiichi nuclear power plant. A refrigeration system (pictured) is creating a 30-metre deep, 1.5-kilometre-long wall of frozen ground that aims to stop groundwater from flowing under the plant and carrying radioactive isotopes into the sea.