LIVE PULSE
4.0 Anthropic CEO Amodei calls for slower AI development and shared safety rules — 11 src2.2 Agility Robotics unveils Digit 5 humanoid for warehouses and factories — 2 src2.0 Apple ships rebuilt Siri with Google Gemini, but not in the EU — 2 src1.8 Siri AI in macOS 27 Golden Gate: FAQ, Germany availability, privacy questions — 2 src1.4 Sam Altman says OpenAI will not go public in 2026, citing AI safety concerns — 5 src1.1 OpenAI contractors review real ChatGPT conversations to rate responses, report says — 2 src1.1 Anthropic data retention policy prompts firms to limit Claude use for sensitive work — 1 src1.1 Study examines issue bias in LLMs used as writing assistants before Swedish 2026 election — 1 src1.1 Study Audits Misalignment in Multi-Modal World Models — 1 src1.1 Retrieval-Grounded Reasoning Approach Proposed for Universal Multimodal Embeddings — 1 src4.0 Anthropic CEO Amodei calls for slower AI development and shared safety rules — 11 src2.2 Agility Robotics unveils Digit 5 humanoid for warehouses and factories — 2 src2.0 Apple ships rebuilt Siri with Google Gemini, but not in the EU — 2 src1.8 Siri AI in macOS 27 Golden Gate: FAQ, Germany availability, privacy questions — 2 src1.4 Sam Altman says OpenAI will not go public in 2026, citing AI safety concerns — 5 src1.1 OpenAI contractors review real ChatGPT conversations to rate responses, report says — 2 src1.1 Anthropic data retention policy prompts firms to limit Claude use for sensitive work — 1 src1.1 Study examines issue bias in LLMs used as writing assistants before Swedish 2026 election — 1 src1.1 Study Audits Misalignment in Multi-Modal World Models — 1 src1.1 Retrieval-Grounded Reasoning Approach Proposed for Universal Multimodal Embeddings — 1 src
HEATPULSEAI MAGAZINES
FLIP · FOLLOW · SAVE
Log inSign up

Generative models

topic6 events
papersTODAY 04:00 UTC

PIVOT Method Uses Physics Cues to Detect AI-Generated Audio-Video

A new arXiv preprint proposes PIVOT, a detection approach that checks whether AI-generated audio and video follow real-world physical behavior rather than relying on visual artifacts. The authors argue that as generative models improve, common artifact-based detectors become less reliable, so physical consistency offers a more durable signal. The abstract is truncated, so full details on the method and evaluation are not available here.

deepfakes/detection/research/multimodal
papersTODAY 04:00 UTC

IsingFormer: Learned Proposals Augment Parallel Tempering for MCMC

A new arXiv paper proposes adding a global proposal move to parallel tempering, in which a learned model suggests state changes across finite temperatures. The aim is to improve mixing in Markov chain Monte Carlo sampling and optimization, where generative models have shown promise but remain difficult to integrate with standard MCMC. The authors present this as a way to combine learned proposals with established tempering methods rather than replace them.

mcmc/sampling/generative-models/research
papersSEP 12 04:00 UTC

Information-Theoretic Framework Unifies Generalization Bounds for VAEs and Diffusion Models

A new arXiv paper derives generalization guarantees for both variational autoencoders and diffusion models within a single information-theoretic framework. The analysis exploits the encoder-generator structure shared by the two model families, which earlier theoretical work had largely treated separately. The authors report bounds that clarify how the shared architecture affects performance on unseen data.

research/diffusion/generative-models/theory
papersSEP 12 04:00 UTC

Benchmark and Method Proposed for Think-with-Video Reasoning in Generative Models

A new arXiv paper argues that while video generation models now produce convincing and temporally consistent output, it is unclear whether they can reason through video by following symbolic rules, obeying physics, and working toward defined goals. The authors introduce a benchmark for measuring this think-with-video ability and propose an approach for improving it. The work is listed under the cs.AI cross-submission category.

video-generation/benchmarks/reasoning/arxiv
papersSEP 10 04:00 UTC

Paper introduces latent bridge matching for albedo estimation in intrinsic image decomposition

A new arXiv preprint presents a latent bridge matching technique for estimating albedo, the reflectance component separated from lighting in intrinsic image decomposition. The authors note that generative approaches to this task have struggled with weak physical plausibility and computationally heavy inference, and their method is designed to tackle both shortcomings. The work is cross-listed under arXiv's artificial intelligence category.

computer-vision/generative-models/image-decomposition
papersSEP 10 04:00 UTC

Jump-Diffusion Framework Introduced for Generating Irregularly Sampled Time Series

A research paper presents a method for training generative models on continuous-time data that is recorded unevenly and out of sync across sources. The approach builds on generator matching and can represent trajectories with sudden jumps rather than only smooth paths, backed by closed-form expressions for diffusion components. It could be useful in domains where measurements arrive at irregular intervals, such as healthcare monitoring or sensor networks.

time-series/generative-models/diffusion/research