Tue 16 Dec 2025 08.00 ESTLast modified on Tue 16 Dec 2025 08.02 EST The sign outside Tom Hermes's farmyard in Perkins Township in Ohio, a short drive south of the shores of Lake Erie, proudly claims that his family have farmed the land here since 1900. Today, he raises 130 head of cattle and grows corn, wheat, grass and soybeans on 1,200 acres of land. For his family, his animals and wider business, water is life. So when, in May 2024, the Texas-based Aligned Data Centers broke ground on its NEO-01, four-building, 200,000 sq ft data center on a brownfield site that abuts farmland that Hermes rents, he was concerned. "We have city water here. That's going to reduce the pressure if they are sucking all the water," he says of the data center.

Over the course of the year, we've had the pleasure of working with many talented researchers from across the globe. As 2025 draws to a close, we take a look back at some of the excellent blog posts from our contributors. This work contributes to the field of explainable AI by developing a novel neural network that can be directly transformed into logic. The authors explore the tensions between creators and AI-generated content through a survey of 459 artists. Find out more about work presented at ECAI on generating a comprehensive biomedical knowledge graph question answering dataset.

Musicians are deeply concerned about AI. So why are the major labels embracing it? Companies such as Udio, Suno and Klay will let you use AI to make new music based on existing artists' work. T his was the year that AI-generated music went from jokey curiosity to mainstream force. Velvet Sundown, a wholly AI act, generated millions of streams; AI-created tracks topped Spotify's viral chart and one of the US Billboard country charts; AI "artist" Xania Monet "signed" a record deal. BBC Introducing is usually a platform for flesh-and-blood artists trying to make it big, but an AI-generated song by Papi Lamour was recently played on the West Midlands show.

Quantum navigation could solve the military's GPS jamming problem The rise of GPS vulnerability is putting more resilient, atom-based navigational tools on the map. The Royal Navy partnered with Infleqtion to test a quantum clock on the uncrewed submarine XV Excalibur. In late September, a Spanish military plane carrying the country's defense minister to a base in Lithuania was reportedly the subject of a kind of attack --not by a rocket or anti-aircraft rounds, but by radio transmissions that jammed its GPS system. The flight landed safely, but it was one of thousands that have been affected by a far-reaching Russian campaign of GPS interference since the 2022 invasion of Ukraine. The growing inconvenience to air traffic and risk of a real disaster have highlighted the vulnerability of GPS and focused attention on more secure ways for planes to navigate the gauntlet of jamming and spoofing, the term for tricking a GPS receiver into thinking it's somewhere else. US military contractors are rolling out new GPS satellites that use stronger, cleverer signals, and engineers are working on providing better navigation information based on other sources, like cellular transmissions and visual data.

A campaign fronted by popstars including Elton John and Dua Lipa to protect artists' works from being mined to train AI models without consent has received a boost after almost every respondent to a government consultation backed their case. Ministers subsequently dropped that preference in the face of a backlash. Liz Kendall, the secretary of state for science, innovation and technology, told parliament on Monday there was "no clear consensus" on the issue and the government would "take the time to get this right", and promised to make policy proposals by 18 March 2026. "This means keeping the UK at the cutting edge of science and technology so UK citizens can benefit from major breakthroughs, transformative innovation and greater prosperity. "It also means continuing to support our creative industries, which make a huge economic contribution, shape our national identity and give us a unique position on the world stage."

Ben & Jerry's row deepens as three board members removed Three members of Ben & Jerry's independent board will no longer be eligible to serve in their roles, after the ice cream company introduced a new set of governance practices. These include a nine-year limit set on board members' terms. Chair Anuradha Mittal, who earlier said she had no plans to resign under pressure, is among those affected. The move was criticised by the company's co-founder Ben Cohen, who called it a blatant power grab designed to strip the board of legal authority and independence. His remarks are the latest in a long-running row between Ben and Jerry's and its owner over the Cherry Garcia maker's social activism and the continued independence of its board.

Three years after ChatGPT made OpenAI the leader in artificial intelligence and a household name, rivals have closed the gap and some investors are wondering if the sensation has the wherewithal to stay dominant. Investor Michael Burry, made famous in the film The Big Short, recently likened OpenAI to Netscape, which ruled the web-browser market in the mid-1990s only to lose to Microsoft's Internet Explorer. OpenAI is the next Netscape, doomed and hemorrhaging cash, Burry said recently in a post on X, formerly Twitter. In a time of both misinformation and too much information, quality journalism is more crucial than ever. By subscribing, you can help us get the story right.

UK launches taskforce to'break down barriers' for women in technology The government has launched a new taskforce it says will help women enter, stay and lead in the UK tech sector. Led by technology secretary Liz Kendall, it will see female leaders from tech companies and organisations advise the government on how to boost diversity and economic growth in the industry. BCS, the Chartered Institute for IT, recently suggested women accounted for only 22% of those working in IT specialist roles in the UK. Ms Kendall said the Women in Tech group would break down the barriers that still hold too many people back. When women are inspired to take on a role in tech and have a seat at the table, the sector can make more representative decisions, build products that serve everyone, she said.

Data attribution aims to explain model predictions by estimating how they would change if certain training points were removed, and is used in a wide range of applications, from interpretability and credit assignment to unlearning and privacy. Even in the relatively simple case of linear regressions, existing mathematical analyses of leading data attribution methods such as Influence Functions (IF) and single Newton Step (NS) remain limited in two key ways. First, they rely on global strong convexity assumptions which are often not satisfied in practice. Second, the resulting bounds scale very poorly with the number of parameters ($d$) and the number of samples removed ($k$). As a result, these analyses are not tight enough to answer fundamental questions such as "what is the asymptotic scaling of the errors of each method?" or "which of these methods is more accurate for a given dataset?" In this paper, we introduce a new analysis of the NS and IF data attribution methods for convex learning problems. To the best of our knowledge, this is the first analysis of these questions that does not assume global strong convexity and also the first explanation of [KATL19] and [RH25a]'s observation that NS data attribution is often more accurate than IF. We prove that for sufficiently well-behaved logistic regression, our bounds are asymptotically tight up to poly-logarithmic factors, yielding scaling laws for the errors in the average-case sample removals. \[ \mathbb{E}_{T \subseteq [n],\, |T| = k} \bigl[ \|\hatθ_T - \hatθ_T^{\mathrm{NS}}\|_2 \bigr] = \widetildeΘ\!\left(\frac{k d}{n^2}\right), \qquad \mathbb{E}_{T \subseteq [n],\, |T| = k} \bigl[ \|\hatθ_T^{\mathrm{NS}} - \hatθ_T^{\mathrm{IF}}\|_2 \bigr] = \widetildeΘ\!\left( \frac{(k + d)\sqrt{k d}}{n^2} \right). \]

Proper quantification of predictive uncertainty is essential for the use of machine learning in safety-critical applications. V arious uncertainty measures have been proposed for this purpose, typically claiming superiority over other measures. In this paper, we argue that there is no single best measure. Instead, uncertainty quantification should be tailored to the specific application. To this end, we use a flexible family of uncertainty measures that distinguishes between total, aleatoric, and epistemic uncertainty of second-order distributions. These measures can be instantiated with specific loss functions, so-called proper scoring rules, to control their characteristics, and we show that different characteristics are useful for different tasks. In particular, we show that, for the task of selective prediction, the scoring rule should ideally match the task loss. On the other hand, for out-of-distribution detection, our results confirm that mutual information, a widely used measure of epistemic uncertainty, performs best. Furthermore, in an active learning setting, epistemic uncertainty based on zero-one loss is shown to consistently outperform other uncertainty measures.