Russia's Putin vows retaliation after accusing Ukraine of hitting student dormitory Russian President Vladimir Putin has promised retaliation after accusing Ukraine of carrying out a deadly attack on a student dormitory in an occupied part of eastern Ukraine. Six people were killed and 39 injured in the overnight strike in the town of Starobilsk, Luhansk region, Putin said. Another 15 people were missing. Ukraine's military said it hit the headquarters of Russia's elite Rubicon drone military unit in Starobilsk. It did not say whether it was the same building as the one identified by Russia.

London Mayor Sir Sadiq Khan has been accused of putting politics over public safety after blocking a £50m contract between the Metropolitan Police and US tech firm Palantir. Scotland Yard had been in talks to use the company's artificial intelligence to speed up criminal investigations. Palantir's UK chief executive Louis Mosley also said the decision would give hostile states and criminals an advantage. The Met has previously warned it will have to cut officer numbers if the deal does not proceed. Palantir, founded by tech billionaire Peter Thiel, a prominent donor to US President Donald Trump, already holds contracts with other UK public sector bodies.

The boss of Standard Chartered has apologised after describing employees whose jobs are vulnerable to being replaced by Artificial Intelligence (AI) as lower value human capital. Discussing how automation was likely to lead to thousands of job cuts at the bank at a recent conference, Bill Winters said it wasn't about cost cutting but replacing, in some cases, lower value, human capital, with the financial capital and the investment capital that we're putting in. He later sought to contextualise the remarks via LinkedIn and said he was sorry for his wording, which had caused upset to some colleagues. He said he was committed to helping staff cope with the accelerating pace of change. The rise of AI tools has led to predictions of huge job losses, particularly for tech workers and graduates.

Sadiq Khan's office blocked Palantir's deal with the Met police, saying there had been a'clear and serious breach' of procurement rules. Sadiq Khan's office blocked Palantir's deal with the Met police, saying there had been a'clear and serious breach' of procurement rules. London mayor accused of'putting politics above public safety' for rejecting deal to use AI in intelligence analysis Fri 22 May 2026 09.45 EDTLast modified on Fri 22 May 2026 09.55 EDT Palantir has accused Sadiq Khan of "putting politics above public safety" after the London mayor blocked its £50m contract with the Metropolitan police in a move that has also led to tensions inside Labour over its involvement with the US tech company. Louis Mosley, who heads Palantir in the UK and Europe, accused Khan of politicising procurement after he rejected a two-year deal for Scotland Yard to use AI to process intelligence in criminal investigations, as first revealed by the Guardian. Mosley said: "What Londoners value is not being mugged, not being raped by a serving police officer." The Met had planned to hire Palantir, which was co-founded by the Trump-supporting tech billionaire Peter Thiel, to automate aspects of investigations.

'At a loss for words' - Destiny 2 fans react to news support is ending The makers of Destiny 2 will stop releasing content updates for the game, effectively drawing to a close one of the industry's longest-running live-service eras. Bungie said the influential online shooter - which has attracted millions of players - will be getting its final update on 9 June, though it will remain playable beyond that. It follows months of fan speculation on the game's future following delays, falling player numbers and the release of Bungie's new shooter, Marathon. UK-based Destiny content creator My name is Byf posted: Saying goodbye like this is more painful than I can fathom, adding: I can only hope the road doesn't end here for good. For fans, the announcement may not have come as a complete surprise, but it was still met with shock and sadness. It's been my entire adult life, said Destiny YouTuber Datto in an emotional video update after the news was released.

Forget the scrap for survival between Tottenham and West Ham, the real battle on the final day of the Premier League season is humans versus machines. The outcome of the BBC predictions title race has gone to the wire, and BBC Sport football expert Chris Sutton is the only one who can stop AI - Microsoft chatbot Copilot - from taking the glory. With one round of fixtures to go, they are level on outright wins - which is the decisive metric. Sutton is in second place because he has managed fewer tied victories, so he needs another outright win on Sunday to finish top of the table. The BBC readers and the guests have put up a pretty pathetic fight, said Sutton. It is just down to me now.

Tune into live broadcasts from your Windows desktop with Trdo, a free and open-source application. I may be old-fashioned, but I prefer actual radio stations to Spotify's algorithms. The best human DJs find music I'd never seek out, and that even the best recommendation system would never point out to me. Even better: If you're good at finding community and public radio stations that appeal to your tastes, there are no commercials. I've found several of my favorite bands in the past few years listening to radio stations like KEXP, Indie XFM, and the various stations offered by SomaFM . It's simple to listen to such stations in your browser, but leaving a tab open just for the radio annoys me.

Fans react to Stephen Colbert's final show Fans lined up outside the Ed Sullivan Theater one last time for "The Late Show" as Stephen Colbert prepared to host his final episode, marking the end of the franchise's 33-year run on CBS. The network announced last summer that the show would end, calling it "purely a financial decision against a challenging backdrop in late night. Colbert took on the show on September 2015 after David Letterman retired from the role he'd held for 22 years. Rescuers removed the vehicle from near the shoreline and police arrested the driver, who was attempting to use the vehicle's wade mode. As part of the ongoing investigation into the crash that killed 14 people, officials released footage showing the engine detaching during takeoff.

Physical computing systems provide a promising route toward hardware-native machine learning, but their computational capabilities remain difficult to characterize in a principled, task-independent, and data-efficient way. We extend the Information Processing Capacity (IPC) framework to stationary physical computing systems and establish several fundamental results: individual capacities are bounded between zero and one, their sum over a complete basis is bounded by the number of readouts, and noise strictly reduces this bound. We address the finite-sample estimation of IPC and derive the asymptotic form of the systematic positive bias affecting naive estimators. Building on these results, we introduce data-efficient estimation methods based on Richardson extrapolation and Sobol quasi-random sampling. We validate the framework experimentally using a photonic computing system based on picosecond laser pulses propagating through a nonlinear optical fibre. By varying the laser power and fibre length, we observe systematic shifts of the IPC distribution toward higher-order nonlinear capacities induced by the Kerr effect. Finally, we demonstrate that the total IPC strongly correlates with performance on benchmark machine-learning tasks and provides a reliable estimate of the effective dimensionality of the system. These results establish IPC as a practical bridge between the intrinsic dynamics of physical computing systems and their machine-learning performance.

We introduce FLASH-MAX, a shallow, exact-by-construction neural network architecture for predicting homogeneous electromagnetic fields from sparse pointwise observations. Each hidden neuron represents a separate exact solution to Maxwell's equations, so that the network satisfies the governing equations symbolically by construction and can be trained end-to-end from sparse data within seconds. We prove a universal approximation result showing that this exact model class remains universal on arbitrary domains. FLASH-MAX reaches sub-1% relative validation error from about 1K sparse pointwise observations in seconds, all while maintaining a zero PDE residual, and keeps single-digit errors even for only 100 observations sampled from 3D space. These results suggest that moving governing structure from the loss into the hypothesis class can dramatically improve the trade-off between precision and optimization speed in scientific machine learning.