Apple has published a new study in its Machine Learning Research blog, called EMBridge: Enhancing Gesture Generalization from EMG Signals through Cross-Modal Representation Learning. This study will be presented at the ICLR 2026 Conference in April. In it, the researchers explain how they trained an AI model to recognize hand gestures, even when those specific hand gestures weren’t part of its original dataset. To achieve this, they developed EMBridge, “a cross-modal representation learning framework that bridges the modality gap between EMG and pose.” EMG, or Electromyography, measures the el
Apple trained an AI to recognize hand gestures from sensor data - 9to5MacEMBridge was the researchers’ way to bridge the gap between real EMG muscle signals and structured hand pose data. Trained using a cross-modal framework, the model was first pre-trained on EMG and hand pose data separately. Then, the researchers aligned the two representations so the EMG encoder could learn from the pose encoder. This allowed EMBridge to learn to recognize gesture patterns from EMG signals. Once that was done, they trained the system using masked pose reconstruction, hiding parts of the pose data and asking the model to reconstruct them using only the information extracted fro
Apple trained an AI to recognize hand gestures from sensor data - 9to5Mac… Importantly, it did so with only 40% of the training data. One important limitation noted in the paper is that the model relies on datasets containing both EMG signals and synchronized hand pose data. This means its training still depends on specialized datasets that can be difficult to collect. …
3 million dating app photos used for AI training before FTC privacy enforcement Ben Lovejoy | Apr 21 2026 - 5:46 am PT | Apr 21 2026 - 5:46 am PT Around 3 million user photos from dating app OkCupid were used for AI training by a third-party company in direct contravention of the company’s own priv… …
… In addition to Apple, the plaintiffs have filed proposed class action suits against Amazon and OpenAI, alleging that both companies also used the Panda-70M dataset in their own AI model training processes. …
… If The Pile rings a bell to you, that’s likely because it surfaced in a different AI training accusation in 2024 , involving YouTube videos. …
… The contents of those documents could potentially also be used as training material. Opting out of AI data collection Fortunately, all major AI apps provide the ability to opt out of using your data as training material, and for most it’s both quick and easy to do. …
… But in this case, upgrading to any of the paid plans doesn’t result in any less data being collected. Many AI chatbots use your questions as training data for their models, so you should always assume that your interactions are not private. …
… Apple will enable Nvidia’s confidential compute feature that encrypts data as it’s being processed on the chips. The Nvidia Blackwell B200 is one of Nvidia’s flagship data center GPUs for large-scale AI training and inference. …
… They randomly sampled 50,000 images from two training datasets, PixMoCap and DenseFusion-4V-100K. …
… Training LiTo To train the model, the researchers selected thousands of objects rendered from 150 different viewing angles, and 3 lighting conditions. …
… Health data sources such as electronic health records or wearables can be disconnected, and information is never used to train AI models or sold to third parties. Read more in our privacy policy. Perplexity isn’t the first AI firm to connect to Apple Health. …