Japan is one of many locations in Asia where Chinese companies access American AI chips -- by renting hardware that's owned by foreign firms and installed in overseas data centers. Taiwan prosecutors suspect that three individuals successfully smuggled at least one shipment of Nvidia artificial intelligence chips to China after first exporting them to Japan, people familiar with the matter said. The trio was detained last week by Taiwan's Keelung District Prosecutors Office for allegedly falsifying documents related to exports of Super Micro Computer servers containing advanced Nvidia chips, which the U.S. has barred from sale to China without a license from Washington. The move marked the island democracy's first public crackdown on AI chip diversion after years of pressure from the U.S. to take a more active role in curtailing China's tech access. When Taiwan authorities apprehended the three defendants -- who've now been officially detained -- they also seized about 50 servers for which they accuse the trio of preparing fraudulent export documents. But at least one shipment had already gone through Taiwan customs, according to the people familiar with the matter, who requested anonymity to speak about an ongoing criminal investigation.

The LAPD has long turned to wealthy boosters to supplement its multi-billion dollar annual budget, but the influx of private money is prompting concerns.

Utilizing artificial intelligence and robots -- and more specifically humanoids -- is crucial in making up for Japan's labor shortage. This was the dominant talking point at the Humanoids Summit on Thursday when the two-day event kicked off in Tokyo. Hosted by a California-based robotics company of the same name, it is the first time the summit, which was previously held in Silicon Valley and London, is being held in Asia. It is expected to draw 2,000 attendees from 30 countries and 300 companies, according to the organizers. Japan was chosen for its "foundational role in the global robotics ecosystem for decades," said Terence Bennett, executive director of the Bay Area Robotics Association, in his opening remarks.

Illinois Lawmakers Just Passed America's Strongest AI Safety Bill The bill requires companies like OpenAI, Anthropic, and Google to have third parties confirm they're following safety standards. The Illinois House of Representatives passed a bill on Wednesday requiring frontier AI labs like OpenAI, Anthropic, and Google DeepMind to have their safety practices audited by a third party. If signed into law, AI safety experts tell WIRED, it would be the nation's leading check on the power of major AI companies . The bill, SB 315, now heads to governor JB Pritzker's desk. In a post on social media on Wednesday, Pritzker said he plans to sign the bill, citing a need to hold Big Tech accountable.

Card payment fraud detection is usually framed as a supervised classification problem. Although this approach has generated practical progress, improvement has remained incremental despite major advances in model architecture. We argue that this is not mainly a failure of function approximation or optimization, but a consequence of structural information impairments inherent to the payment ecosystem. We formalize card authorization as a sequential decision problem with delayed, censored, corrupted, and counterfactually missing feedback. We derive a minimax regret lower bound showing that these impairments enter multiplicatively in the denominator of the achievable learning rate. The bound implies that improving issuer reporting quality or reducing censorship can yield larger reductions in the regret floor than increasing model complexity. We also show that heterogeneity across issuers worsens learnability beyond what average impairment rates suggest. The paper contributes a theory of why fraud detection in payment networks is fundamentally harder than in standard online learning settings, identifies ecosystem information quality as the key bottleneck, and provides a theoretical basis for prioritizing investments in reporting infrastructure, dispute process quality, and selective exploration. The paper is theory-first and does not rely on proprietary transaction data.

Fine-tuning a pretrained language model on a curated dataset can produce spurious correlations between the fine-tuning task and unintended latent factors -- such as misaligned personas or political slant -- that the curation procedure has entangled with the task. The model can latch onto these spurious correlations, leading to bias and reduced out-of-distribution generalisation. We prove that under reasonable assumptions on task complexity and the spurious correlation, such latent factors can be identified, without supervision, from the weights of a naive LoRA fine-tune. Existing approaches to removing bias, such as activation steering, remove identified factors from residual-stream activations, either at inference or during training. We argue, however, that the goal should be to remove the spurious correlation, not the latent factor itself, as the pretrained model may rely on it for genuine task signal. To enable this, we propose GRASP, GRadient projection of Associated Spurious Patterns, which prevents the model from acquiring new reliance on the identified latent factor while preserving any pretrained content along it. We validate on three fine-tuning tasks. The first two involve emergent misalignment, where fine-tuning on a narrow task -- in our case, writing insecure code and giving bad medical advice -- leads to misaligned responses on unrelated topics. Here our method completely removes misalignment in the insecure code case and reduces them by ~5x in the bad medical advice case, beating all baselines in the trade-off between misalignment-reduction and task-preservation. The last is a novel political-bias experiment, where fine-tuning on right-skewed Reddit financial-advice data causes political-lean drift on unrelated topics. Here our method reduces drift by more than half, while improving financial task performance, beating all baselines.

We consider the problem of learning a counterfactually fair regressor. We adopt a causal uncertainty view in which counterfactual fairness is defined with resampled noise. We focus on obtaining theoretical fairness guarantees for a new post-processing estimator. We begin by showing that counterfactual fairness is equivalent to satisfying demographic parity conditional on the latent variable. This allows us to provide a closed-form expression of the optimal fair regressor via a barycentric quantile map. In order to handle continuous latent variables, we propose a discretized post-processing method. Then, under mild regularity assumptions, we prove high-probability finite-sample fairness guarantees for our estimator, providing an unfairness decay at rate $\tilde O(n^{-1/3})$, and establishing a matching risk bound of order $\tilde O(n^{-1/3})$. We provide a matching lower bound on the excess risk of almost fair predictions. Finally, we extend our results to the setting of relaxed counterfactual fairness. We validate our approach on real-world and synthetic data.

We analysed thousands of Trump's posts - here's what we found In 2026, Donald Trump's use of social media has escalated. The BBC sifted through thousands of posts on his platform Truth Social to analyse what the President has been saying and when. What was the busiest day? When are the busiest hours? What type of content does President Trump share?

According to federal prosecutors, Michele Spagnuolo made more than $1 million on the prediction market platform using confidential information about Google Search traffic. A Google security engineer has been charged with crimes stemming from allegedly placing trades on Polymarket using confidential internal information from the tech giant. Michele Spagnuolo, a 36-year-old Italian citizen, was arrested this morning in New York, as first reported by ABC News. Spagnuolo is charged with one count each of commodities fraud, wire fraud, and money laundering. He has worked at Google since 2014 and was based out of the company's Zurich, Switzerland, offices.

Russian lawmakers have passed a bill to allow trained bank employees to shoot down Ukrainian drones amid an increase in the number of attacks. The draft legislation, which would see banks across Russia install electronic jamming systems while selected employees would shoot down incoming unmanned aircraft, passed in its third and final reading in the lower house Duma on Tuesday, according to the state-run TASS news agency. The bill says the legislation is needed to protect Bank of Russia facilities, including those located in the new constituent entities of the Russian Federation - referring to the four eastern Ukrainian regions that Moscow has announced it has annexed despite not controlling them fully - amid the increasing number of sabotage and terrorist attacks. Under the plan, banks would finance the installation of the equipment on their premises. With banks in almost every town, their incorporation into Russia's air defences could help expand its cover.