The day's stories
Product LaunchRunway launches AI model router as generative media gets crowdedRunway has introduced the Media Router, a new tool aimed at developers working with generative media. This innovative solution allows users to choose from a varTechCrunch AIJul 23· 63%FundingAI chip startup Etched defies skeptics, hits $10.3B valuation from big-name investorsEtched, an AI chip startup founded by Harvard dropouts, has successfully raised $300 million in a Series C funding round, propelling its valuation to an impressTechCrunch AIJul 23· 60%Product LaunchNvidia is sending GPUs to the moonNvidia is making headlines with its latest initiative to send its Jetson chips to the moon, which will be utilized in Lunar Outpost's upcoming moon rover. This TechCrunch AIJul 23· 59%Product LaunchOpenAI makes ChatGPT Health available to all US usersOpenAI has officially launched its ChatGPT Health feature for all users in the United States aged 18 and over. This new functionality allows users to integrate TechCrunch AIJul 23· 59%ResearchExperts say exploiting Anthropic's Fable isn't how Kimi K3 got so goodRecent discussions have emerged regarding the development of Kimi K3, an advanced open-weight language model by the Chinese company Moonshot. White House sciencTechCrunch AIJul 23· 58%Product LaunchAMD takes on Nvidia with its Helios AI rack-scale systemAMD has officially launched its Helios AI rack-scale system, positioning itself as a competitor to Nvidia in the high-performance computing space tailored for ATechCrunch AIJul 23· 58%Product LaunchMeta launched a new AI optimism ad set to a song about human extinctionMeta has unveiled a new advertisement aimed at promoting optimism around artificial intelligence. The ad showcases positive visuals and a voiceover that seeks tTechCrunch AIJul 23· 58%Product LaunchBringing Nunchaku 4-bit Diffusion Inference to DiffusersHugging Face has announced the integration of the Nunchaku 4-bit Diffusion Inference engine into its Diffusers library. This advancement allows users to load moHugging Face BlogJul 23· 57%FundingServiceNow bets $40 million on Indian banking software specialist to expand its financial services pushServiceNow has made a strategic investment of $40 million in BusinessNext, an Indian banking software specialist, acquiring a 5% stake in the company. This moveTechCrunch AIJul 23· 57%FundingAegisAI, founded by former Google security execs, lands $36M to stop AI-driven spear phishingAegisAI, a new startup established by ex-Google security executives, has successfully secured $36 million in funding aimed at tackling the rising threat of AI-dTechCrunch AIJul 23· 57%Product LaunchAnthropic updates Claude voice mode with more capable modelsAnthropic has rolled out significant updates to its Claude voice mode, introducing three new models: Opus, Sonnet, and Haiku. These enhancements are designed toTechCrunch AIJul 23· 57%Open SourceOpenAI won't let some customers export their chats, but this tool willOpenAI has restricted its Business and Enterprise users from easily exporting their chat data, which has raised concerns among users about data ownership and acThe Register AIJul 23· 57%OtherIBM insists AI didn't kill software deals, just delayed themIBM has addressed concerns regarding the impact of AI on its software deals, clarifying that the delays experienced were not indicative of a long-term downturn The Register AIJul 23· 56%Product LaunchGoogle's Gemini nears billion-user milestoneGoogle's AI assistant, Gemini, is on the verge of reaching a significant milestone with nearly a billion users. During the Q2 2026 earnings call, it was reporteTechCrunch AIJul 23· 56%BenchmarkAir Quality Arena: A Large-Scale Multi-Region Ground Monitoring Dataset and Benchmark for Air Quality Forecasting with Time-Series Foundation ModelsThe Air Quality Arena (AQA) has been introduced as a large-scale dataset and benchmark aimed at enhancing air quality forecasting. This initiative aggregates daarXiv cs.LGJul 23· 64%ResearchSemantic Cooperative Games for Contribution Attribution in LLM-Based Multi-Agent SystemsRecent research has introduced a novel framework known as Semantic Cooperative Games (SCG) aimed at enhancing contribution attribution in large language model (arXiv cs.AIJul 23· 64%ResearchPEARL: Solver-in-the-Loop Interactive Optimization Modeling from Natural LanguageResearchers have unveiled PEARL, a novel system designed to enhance interactive optimization modeling by integrating solver feedback. This innovative approach aarXiv cs.AIJul 23· 63%BenchmarkWhen JSON Is Not Enough: Semantic Reliability of Schema-Constrained LLM Ordering AgentsA recent study introduced OrderBench, a benchmark designed to evaluate the reliability of schema-constrained large language model (LLM) ordering agents. The resarXiv cs.AIJul 23· 63%BenchmarkFindStatBench: Evaluating Large Language Models on Combinatorial Code SynthesisThe introduction of FindStatBench marks a significant advancement in the evaluation of large language models, specifically in the realm of combinatorial code syarXiv cs.AIJul 23· 63%ResearchChallenges of Explainability in Continual Learning for Time Series ForecastingRecent research has delved into the complexities of explainability within continual learning frameworks specifically for time series forecasting. The study higharXiv cs.LGJul 23· 63%ResearchProbabilistic Concept-Aware Steering for Trustworthy LLM InferenceA new framework called Probabilistic Concept-Aware Steering (PCS) has been introduced to enhance the inference capabilities of large language models (LLMs). ThiarXiv cs.AIJul 23· 63%ResearchS2T-RLHF: Hierarchical Credit Assignment for Stable Preference-Based RLHFA new framework called S2T-RLHF has been introduced to improve the stability of reinforcement learning from human feedback (RLHF). This innovative approach shifarXiv cs.AIJul 23· 63%ResearchCross-Dialect Generalization Without Retraining: Benchmarks and Evaluation of Schema-Derived Constrained Decoding for MLIRA recent study has been published that explores the capabilities of Multi-Level Intermediate Representation (MLIR) in the context of natural-language-to-MLIR trarXiv cs.AIJul 23· 63%ResearchBayesian Wind Tunnels for Model SelectionRecent research has introduced the concept of Bayesian wind tunnels for model selection, highlighting the effectiveness of a 2.8M-parameter transformer in identarXiv cs.LGJul 23· 62%ResearchMUX: Continuous Reasoning via Multiplexed TokensResearchers have unveiled a new approach called MUX, which focuses on enhancing reasoning capabilities in language models by utilizing continuous multiplexed toarXiv cs.AIJul 23· 62%ResearchSTN-TGAT: Top-K Portfolio Construction via Prior-Guided Graph Attention with Learnable Soft-Threshold SparsificationThe recent paper on STN-TGAT introduces an innovative approach to stock ranking and portfolio construction by integrating a temporal Transformer with a Graph AtarXiv cs.LGJul 23· 61%BenchmarkCruiseBench: A Real-Flight-Aligned N-CMAPSS Benchmark for Engine RUL PredictionThe introduction of CruiseBench marks a significant advancement in the benchmarking of remaining useful life (RUL) prediction models for engines. This new bencharXiv cs.LGJul 23· 61%ResearchBuilding Fast, Evaluating Slow: Pipeline Choices Dominate Autointerpretability Score VarianceRecent research has uncovered that the choices made in the evaluation pipeline for sparse autoencoders have a more significant impact on autointerpretability scarXiv cs.LGJul 23· 61%ResearchNeural Operator Surrogates for Two-Dimensional Neutron Flux EstimationRecent research has advanced the field of neutron flux estimation by extending neural operator techniques from one-dimensional to two-dimensional applications. arXiv cs.LGJul 23· 60%ResearchPosition: AI/ML Deepfake Research is Misaligned with AI-Generated Non-Consensual Intimate Imagery (AIG-NCII)A recent paper has raised critical concerns regarding the direction of AI/ML deepfake research, arguing that it is overly focused on epistemic harms while largearXiv cs.AIJul 23· 59%