However, prior work has shown that the data can actually be recovered by the server using so-called gradient inversion attacks.

Large Language Models (LLMs) are increasingly deployed in various applications. As their usage grows, concerns regarding their safety are rising, especially in maintaining harmless responses when faced with malicious instructions.

Gaussian mechanism borrowed from differential privacy.

The tech bros might show more humility in Delhi - but will they make AI any safer? Those who shout the loudest about artificial intelligence tend to be in the West, notably the US and Europe. So it's significant that a gathering of powerful leaders is being held in the Global South, a region of the world that runs the risk of being left behind in the AI race. Tech bosses, politicians, scientists, academics and campaigners are meeting at the AI Impact Summit in India this week for top-level discussions about what the world should be doing to try to marshal the AI revolution in the right direction. At last year's AI Action Summit, as it was then known, an ugly power struggle broke out between some Western countries over who should be in charge.

The hazard ratio from the Cox proportional hazards model is a ubiquitous summary of treatment effect. However, when hazards are non-proportional, the hazard ratio can lose a stable causal interpretation and become study-dependent because it effectively averages time-varying effects with weights determined by follow-up and censoring. We consider the average hazard (AH) as an alternative causal estimand: a population-level person-time event rate that remains well-defined and interpretable without assuming proportional hazards. Although AH can be estimated nonparametrically and regression-style adjustments have been proposed, existing approaches do not provide a general framework for flexible, high-dimensional nuisance estimation with valid sqrt{n} inference. We address this gap by developing a semiparametric, doubly robust framework for covariate-adjusted AH. We establish pathwise differentiability of AH in the nonparametric model, derive its efficient influence function, and construct cross-fitted, debiased estimators that leverage machine learning for nuisance estimation while retaining asymptotically normal, sqrt{n}-consistent inference under mild product-rate conditions. Simulations demonstrate that the proposed estimator achieves small bias and near-nominal confidence-interval coverage across proportional and non-proportional hazards settings, including crossing-hazards regimes where Cox-based summaries can be unstable. We illustrate practical utility in comparative effectiveness research by comparing immunotherapy regimens for advanced melanoma using SEER-Medicare linked data.