Is the war entering a new phase? Ukraine launches major drone attack on Russia's Moscow region, killing two Share Ukraine launches major drone attack on Russia's Moscow region, killing two on social media Ukraine has launched a major drone attack on Moscow, killing two people and setting fire to a warehouse in the capital city, Russian officials say. Moscow Mayor Sergei Sobyanin said on Tuesday that the overnight assault involved 650 drones heading towards the city, most of which were shot down, as both sides intensify long-range strikes on infrastructure before winter. He said seven people had been injured and that 47 houses, four apartments and three cars had been damaged. He said 185 drones were "destroyed or neutralised" over the wider region.

To play this video you need to enable JavaScript in your browser. Major bridges in Ukraine have been the target of Russian strikes over the past four days, as Moscow ramps up attacks in Ukraine's capital city. The Pivnichnyi (Northern) Bridge, has been hit for the second day in a row, while the Pivdennyi (Southern) Bridge was hit multiple times earlier this week. The latest attack came as German Chancellor Friedrich Merz made a visit to Kyiv on Sunday. BBC Ukraine correspondent Dan Johnson is near one of the bridges that was targeted and explains what we know so far.

A major bridge in Kyiv has been targeted by Russian strikes for the second day in a row, coinciding with a surprise visit from German Chancellor Friedrich Merz's to the city. The Pivnichnyi (Northern) Bridge was hit on Sunday morning, a day after it was first struck by drones in an attack that damaged the road surface and led to traffic disruption. Kyiv Mayor Vitali Klitschko said Sunday's strike had also damaged a heating main, adding that there would be a temporary lack of hot water in the north-eastern Desnyanskyi district as a result. Major bridges have been targeted in Kyiv over the past four days as Moscow ramps up attacks on critical infrastructure in the capital. Unverified video footage of Sunday's strike shows cars moving on the bridge as a drone crashes into it before exploding, creating a ball of fire.

Image caption, The Northern Bridge is the second major bridge to be targeted by Russia in recent days, Kyiv's mayor said A Russian strike has hit a second major bridge in Kyiv, Mayor Vitali Klitschko has said, as Russia ramps up attacks on critical infrastructure in the Ukrainian capital. The attack on Saturday morning damaged the road surface and trolleybus overhead wires on the Northern Bridge, Klitschko said. The strike comes after another key bridge in the city, Southern Bridge, has been targeted multiple times by Russian drones since Thursday, with Klitschko earlier warning that such attacks were tearing Kyiv apart. The two bridges link the eastern and western parts of the capital and had not been hit before Thursday, signalling a new stage in Moscow's attacks. Image caption, In the Thursday attack, a Russian drone hit cables of Kyiv's Pivdennyi bridge Writing on Telegram on Saturday, Klitschko said: Hitting the North Bridge.

Image caption, All metro and automobile traffic is now suspended across Kyiv's Pivdennyi bridge Intensified Russian drone and missile strikes are tearing Kyiv apart, the mayor of Ukraine's capital has warned. Vitaliy Klitschko said the city was now in a very dramatic situation as Moscow was battering its critical infrastructure, logistics, housing and other key facilities. He was speaking after Kyiv's defence council held a crisis meeting, following new Russian strikes on a key bridge that caused traffic chaos. The city's data centres have also been recently hit, resulting in internet outages. President Vladimir Putin earlier said Russia was acting in response to Ukraine's strikes on Russian energy facilities, despite the fact that it was Moscow that had first launched such attacks.

Two Russian drones have hit one of Kyiv's main bridges, bringing public transport in the Ukrainian capital to a halt and signalling a new stage in Moscow's attacks. The Pivdennyi or Southern bridge came under fire as a metro train was travelling across. Passengers described how dust from a blast wave entered their carriage through the ventilation system. The Pivdennyi bridge is one of the main arteries across the Dnipro river dividing western and eastern Kyiv. Kyiv's bridges had not been hit before Thursday's strikes - and despite a dramatic explosion close to a main support section, city authorities said preliminary checks did not find any critical damage to the structure.

One of AI's Fiercest Critics Says All the Doom Talk Is'Meant to Distract Us' Timnit Gebru argues that AI companies are stoking fears of extinction to avoid discussing actual harms, like autonomous weapons. Here at WIRED we've been trying to make sense of at least two incidents in the AI industry this week: First, the major fight over a million-dollar math problem and whether OpenAI surreptitiously borrowed the work of other researchers to solve the problem first. A day later, an Anthropic researcher--who had previously worked at OpenAI--very publicly quit his job, citing concerns about how both companies were handling AI safety. Another technical staffer at Anthropic said in response that people at the company "really do earnestly believe AI could kill all humans! I personally think it is >10% within the next decade."

A startup called Mostik has a wild new approach to combining the capabilities of AI models. I recently met with some brilliant Russian mathematicians who showed me a way for artificial intelligence models to communicate via something akin to machine telepathy. The mathematicians work for a startup called Mostik--the Russian word for bridge. It's a nod to the group's approach, which allows different models to interact using the mathematical values found in their weights --the things that determine how a prompt gets turned into an output. In practice, this means the capabilities of a larger model can be fed to a smaller model to ramp up its intelligence much more efficiently.

A driver said he was left perturbed after receiving a fine for crossing a toll bridge more than 270 miles from his home despite insisting he had never been near it. Graham Parsons, from Plymouth, Devon, received an unpaid toll charge for using the Warburton Toll Bridge, which links Cheshire and Greater Manchester. His case is one of a number raised by motorists who have complained about the bridge's payment and enforcement system. Peel Ports said there had been some genuine customer experience issues, but the evidence did not indicate a systemic failure of the system. The bridge previously cost 12p a crossing, but the charge was increased to £1 following refurbishment works in recent years .

Building on a two-parameter Weibull framework for diagnosing transformer weight distributions, we study why the Weibull weight-scale parameter $λ$ grows, overshoots, and then relaxes during AdamW training. We derive a leading-order three-force decomposition of the squared weight norm from the AdamW update: an alignment force measuring the correlation between weights and the adaptive update direction, an injection force from adaptive step magnitude, and a decay force from decoupled weight decay. On self-trained Pythia-70M models with ground-truth optimizer moments, alignment dominates the rise phase, contributing 88-94% of the absolute force budget across four random seeds and remaining robust to super-weight removal. Near saturation, alignment and decay approach balance, explaining the transition from weight-scale growth to relaxation. These force dynamics directly govern the squared-norm component underlying $λ(t)$; the remaining RMS-to-Weibull reconstruction offset is measurable and decomposes into bridge and integration components, totaling approximately 5-6% in densely sampled regions. To extend the analysis to real models where optimizer moments are unavailable, we introduce a spline displacement method that recovers the alignment force from sparse checkpoints with approximately 92-94% accuracy, about twice the naive two-point baseline. We further observe that the peak value of $λ(t)$ varies with training-data coherence in our experiments, suggesting a data-dependent component of weight-scale growth that we leave to a controlled follow-up study. Code and data are available at https://github.com/tiexinding/NPM-Weibull-public.