Things to Do in L.A. Tap to enable a layout that focuses on the article. His students suddenly started getting A's. Did a Google AI tool go too far? Google's Lens tool on Chromebooks can mean it easier for students to cheat with one click, prompting teachers to question how they can maintain academic integrity. Over 70% of teachers worry AI tools are preventing students from developing critical thinking and writing skills.

'It reflects and reinforces a culture where consent and respect for personal boundaries are undermined.' 'It reflects and reinforces a culture where consent and respect for personal boundaries are undermined.' The rise of deepfake pornography in schools: 'One girl was so horrified she vomited' The use of'nudify' apps is becoming more and more prevalent, with hundreds of teachers having seen images created by pupils, often of their peers. He didn't feel this was something he shouldn't be doing. It was in the open and people saw it.

UK share values'most stretched' since 2008, Bank warns The Bank of England has warned of a sharp correction in the value of major tech companies with growing fears of an artificial intelligence (AI) bubble. It said share prices in the UK are close to the most stretched they have been since the 2008 global financial crisis, while equity valuations in the US are reminiscent of those before the dotcom bubble burst. The central bank's financial stability report warned valuations are particularly stretched for companies focused on AI. It said the growth of the sector in the next five years would be fuelled by trillions of dollars of debt, raising financial stability risks if the value of the companies falls. The Bank of England cited industry figures forecasting spending on AI infrastructure could top $5tn (£3.8tn).

What is in the 28-point US plan for Ukraine? 'Ukraine is running out of men, money and time' Can the US get all sides to end the war? Why is Europe opposing Trump's peace plan? Zelenskyy says US peace plan'looks better' with new revisions Here's where things stand on Tuesday, December 2: Russian forces launched a ballistic missile on Ukraine's Dnipro, killing four people and wounding 40 others, according to Ukrainian authorities. Russia claimed the capture of the strategic eastern Ukrainian town of Pokrovsk, the logistics hub that has been under attack for months by Moscow's forces.

U.S. moves to deepen minerals supply chain in AI race with China The U.S. is looking to cut its dependence on China. The U.S. will seek agreements with eight allied nations as part of a fresh effort to strengthen supply chains for the computer chips and critical minerals needed for artificial intelligence technology, according to the top State Department official for economic affairs. The initiative, which builds on efforts dating back to the first administration of President Donald Trump, unfolds as the U.S. looks to cut its dependence on China. It will begin with a meeting at the White House on Dec. 12 between the U.S. and counterparts from Japan, South Korea, Singapore, the Netherlands, the U.K., Israel, the United Arab Emirates and Australia, Jacob Helberg, the undersecretary of state for economic affairs, said in an interview. Helberg, a former adviser at Palantir Technologies, said the summit will focus on reaching agreements across the areas of energy, critical minerals, advanced manufacturing semiconductors, AI infrastructure, and transportation logistics.

Apple thanked John Giannandrea for his efforts. Apple thanked John Giannandrea for his efforts. Siri-us setback: Apple's AI chief steps down as company lags behind rivals Apple's head of artificial intelligence, John Giannandrea, is stepping down from the company. The move comes as the Silicon Valley giant has lagged behind its competitors in rolling out generative AI features, in particular its voice assistant Siri. Apple made the announcement on Monday, thanking Giannandrea for his seven-year tenure at the company.

The shock resignation came for a very specific reason, but the OBR saga will continue with a series of decisions the chancellor will have to make over Richard Hughes' replacement. Firstly the Chancellor will have to find a respected and credible economist to run the OBR. There are several candidates, who might fit the mould of fiercely independent bean counters. The list will be carefully watched by the markets for any departure from the normal model. The problem is that there is some political pressure to do just that.

Safety-critical environments are inherently dynamic. Distribution shifts, emerging vulnerabilities, and evolving requirements demand continuous updates to machine learning models. Yet even benign parameter updates can have unintended consequences, such as catastrophic forgetting in classical models or alignment drift in foundation models. Existing heuristic approaches (e.g., regularization, parameter isolation) can mitigate these effects but cannot certify that updated models continue to satisfy required performance specifications. We address this problem by introducing a framework for provably safe model updates. Our approach first formalizes the problem as computing the largest locally invariant domain (LID): a connected region in parameter space where all points are certified to satisfy a given specification. While exact maximal LID computation is intractable, we show that relaxing the problem to parameterized abstract domains (orthotopes, zonotopes) yields a tractable primal-dual formulation. This enables efficient certification of updates - independent of the data or algorithm used - by projecting them onto the safe domain. Our formulation further allows computation of multiple approximately optimal LIDs, incorporation of regularization-inspired biases, and use of lookahead data buffers. Across continual learning and foundation model fine-tuning benchmarks, our method matches or exceeds heuristic baselines for avoiding forgetting while providing formal safety guarantees.

Data privacy is a critical challenge in modern medical workflows as the adoption of electronic patient records has grown rapidly. Stringent data protection regulations limit access to clinical records for training and integrating machine learning models that have shown promise in improving diagnostic accuracy and personalized care outcomes. Synthetic data offers a promising alternative; however, current generative models either struggle with time-series data or lack formal privacy guaranties. In this paper, we enhance a state-of-the-art time-series generative model to better handle longitudinal clinical data while incorporating quantifiable privacy safeguards. Using real data from chronic kidney disease and ICU patients, we evaluate our method through statistical tests, a Train-on-Synthetic-Test-on-Real (TSTR) setup, and expert clinical review. Our non-private model (Augmented TimeGAN) outperforms transformer- and flow-based models on statistical metrics in several datasets, while our private model (DP-TimeGAN) maintains a mean authenticity of 0.778 on the CKD dataset, outperforming existing state-of-the-art models on the privacy-utility frontier. Both models achieve performance comparable to real data in clinician evaluations, providing robust input data necessary for developing models for complex chronic conditions without compromising data privacy.

Many emerging applications - such as adversarial training, AI alignment, and robust optimization - can be framed as zero-sum games between neural nets, with von Neumann-Nash equilibria (NE) capturing the desirable system behavior. While such games often involve non-convex non-concave objectives, empirical evidence shows that simple gradient methods frequently converge, suggesting a hidden geometric structure. In this paper, we provide a theoretical framework that explains this phenomenon through the lens of hidden convexity and overparameterization. We identify sufficient conditions - spanning initialization, training dynamics, and network width - that guarantee global convergence to a NE in a broad class of non-convex min-max games. To our knowledge, this is the first such result for games that involve two-layer neural networks. Technically, our approach is twofold: (a) we derive a novel path-length bound for the alternating gradient descent-ascent scheme in min-max games; and (b) we show that the reduction from a hidden convex-concave geometry to two-sided Polyak-Łojasiewicz (PŁ) min-max condition hold with high probability under overparameterization, using tools from random matrix theory.