AI coding platform's flaws allow BBC reporter to be hacked The BBC has been shown a significant - and unfixed - cyber-security risk in a popular AI coding platform. Orchids is a so-called vibe-coding tool, meaning people without technical skills can use it to build apps and games by typing a text prompt into a chatbot. Such platforms have exploded in popularity in recent months, and are often heralded as an early example of how various professional services could be done quickly and cheaply by AI. But experts say the ease with which Orchids can be hacked demonstrates the risks of allowing AI bots deep access to our computers in exchange for the convenience of allowing them to carry out tasks autonomously. The BBC has repeatedly asked the company for comment but it has not replied.

Doctor at Jeffrey Epstein's post-mortem says the paedophile was strangled and NOT hanged Married California coffee growing pioneers die in'tragic accident' leaving their three children orphaned Lindsey Vonn's'primary goal is to keep her leg': Knee specialist warns Winter Olympics legend could face amputation or'lifelong consequences' after'motorcycle-style' skiing crash Disturbing Kurt Cobain autopsy details revealed for first time: As new probe claims Nirvana singer's death was a HOMICIDE, here's the evidence that convinced forensic investigators HGTV star was canned after vile video leak... now insiders make bombshell claims about what else she was doing behind the scenes: 'Unhinged' Gwyneth Paltrow's'nepo baby' Apple Martin, 21, reveals what cosmetic procedures she has had done Nashville's hottest couple engulfed by cheating storm as insiders declare: 'It's over' Jill Zarin's replacement REVEALED after racist Bad Bunny rant got her sacked from her TV comeback Scientists are ...

AI researcher says'world is in peril' and quits to study poetry An AI safety researcher has quit US firm Anthropic with a cryptic warning that the world is in peril. In his resignation letter shared on X, Mrinank Sharma told the firm he was leaving amid concerns about AI, bioweapons and the state of the wider world. He said he would instead look to pursue writing and studying poetry, and move back to the UK to become invisible. It comes in the same week that a OpenAI researcher said she had resigned, sharing concerns about the ChatGPT maker's decision to deploy adverts in its chatbot . Anthropic, best known for its Claude chatbot, had released a series of commercials aimed at OpenAI, criticising the company's move to include adverts for some users.

Annealed importance sampling (AIS) and related algorithms are highly effective tools for marginal likelihood estimation, but are not fully differentiable due to the use of Metropolis-Hastings correction steps. Differentiability is a desirable property as it would admit the possibility of optimizing marginal likelihood as an objective using gradient-based methods. To this end, we propose Differentiable AIS (DAIS), a variant of AIS which ensures differentiability by abandoning the Metropolis-Hastings corrections. As a further advantage, DAIS allows for mini-batch gradients. We provide a detailed convergence analysis for Bayesian linear regression which goes beyond previous analyses by explicitly accounting for the sampler not having reached equilibrium.

Technology Internet Japan's favorite beer is in peril Asahi Super Dry's manufacturer is suffering from a major cyberattack. Breakthroughs, discoveries, and DIY tips sent every weekday. Japan is facing a serious beer crisis. The emergency began on Monday, September 29 when the makers of the country's most popular brew Asahi Super Dry announced it had suffered a massive cyberattack resulting in a nationwide "system failure." The immediate fallout included a temporary shutdown of nearly all of Asahi Group's 30 domestic breweries, as well a pause in ordering and shipping across Japan.

Traditional models for pricing catastrophe (CAT) bonds struggle to capture the complex, relational data inherent in these instruments. This paper introduces CATNet, a novel framework that applies a geometric deep learning architecture, the Relational Graph Convolutional Network (R-GCN), to model the CAT bond primary market as a graph, leveraging its underlying network structure for spread prediction. Our analysis reveals that the CAT bond market exhibits the characteristics of a scale-free network, a structure dominated by a few highly connected and influential hubs. CATNet demonstrates high predictive performance, significantly outperforming a strong Random Forest benchmark. The inclusion of topological centrality measures as features provides a further, significant boost in accuracy. Interpretability analysis confirms that these network features are not mere statistical artifacts; they are quantitative proxies for long-held industry intuition regarding issuer reputation, underwriter influence, and peril concentration. This research provides evidence that network connectivity is a key determinant of price, offering a new paradigm for risk assessment and proving that graph-based models can deliver both state-of-the-art accuracy and deeper, quantifiable market insights.

Annealed importance sampling (AIS) and related algorithms are highly effective tools for marginal likelihood estimation, but are not fully differentiable due to the use of Metropolis-Hastings correction steps. Differentiability is a desirable property as it would admit the possibility of optimizing marginal likelihood as an objective using gradient-based methods. To this end, we propose Differentiable AIS (DAIS), a variant of AIS which ensures differentiability by abandoning the Metropolis-Hastings corrections. As a further advantage, DAIS allows for mini-batch gradients. We provide a detailed convergence analysis for Bayesian linear regression which goes beyond previous analyses by explicitly accounting for the sampler not having reached equilibrium.

Annealed importance sampling (AIS) and related algorithms are highly effective tools for marginal likelihood estimation, but are not fully differentiable due to the use of Metropolis-Hastings correction steps. Differentiability is a desirable property as it would admit the possibility of optimizing marginal likelihood as an objective using gradient-based methods. To this end, we propose Differentiable AIS (DAIS), a variant of AIS which ensures differentiability by abandoning the Metropolis-Hastings corrections. As a further advantage, DAIS allows for mini-batch gradients. We provide a detailed convergence analysis for Bayesian linear regression which goes beyond previous analyses by explicitly accounting for the sampler not having reached equilibrium.

"We are going to speak of the future," the Polish writer Stanisław Lem wrote, in "Summa Technologiae," from 1964, a series of essays, mostly on humanity and the evolution of technology. "Yet isn't discoursing about future events a rather inappropriate occupation for those who are lost in the transience of the here and now?" Lem, who died in 2006 at the age of eighty-four, is likely the most widely read writer of science fiction who is not particularly widely read in the United States. His work has been translated into more than forty languages, many millions of copies of his books have been printed, and yet, if I polled a hundred friends, 2.3 of them would know who he was. His best-known work in the U.S. is the 1961 novel "Solaris," and its renown stems mostly from the moody film adaptation by Andrei Tarkovsky. Among Lem's fictional imaginings are a phantomatic generator (a machine that gives its user an extraordinarily vivid vision of an alternate reality), an opton (an electronic device on which one can read books), and a network of computers that contains information on most everything that is known and from which people have a difficult time separating themselves.

Industry 4.0 is enabled through the convergence of information technology (IT) and operational technology (OT) in industrial control systems (ICSs).2 At the core of Industry 4.0 are the cyber-physical systems (CPSs), such as power grids, manufacturing industries, autonomous vehicles, smart healthcare, and so forth connecting physical (OT) and cyber (IT) components through computational and networking capabilities.2 While CPSs facilitate automation and resource optimization, they introduce an expanded attack surface that spans both the cyber and physical domains.1 The evolution of tradecraft, from Stuxnet to Industroyer, on energy and utility infrastructure has shown the repercussions of such attacks on economic, business, and social sectors.5 Securing an operational CPS against potential attack vectors involves evaluating the system's operational behavior and assessing security posture. To take security measures effectively, such assessments must occur without negatively affecting the ongoing operations, be reproducible for further investigation, and cover the system's life cycle.2