A new report from Cobalt finds that support for fully automated pentesting among security professionals has plummeted from 29% to 9% in one year, as automated tools fail to detect critical vulnerabili…
Researchers introduced dynamic belief updates in AI models, boosting scientific discovery by 30.62%. The approach challenges static methods like AutoDiscovery's Bayesian surprise, revealing that 37.5%…
Memory-augmented AI agents using TraceRetain show improved performance in clean environments, but retention strategies yield smaller differences than expected. In noisy settings, TraceRetain-CEM maint…
Google's NotebookLM is adding a feature that converts user notes into 60-second TikTok-style AI videos, available exclusively to AI Ultra and Pro subscribers. The tool aims to make complex information…
OrgForge launched a multi-agent simulation framework that enforces a strict physics-cognition boundary to eliminate AI hallucinations in enterprise corpora, achieving a 0.46 absolute improvement in pr…
A KFF poll of 2,480 US adults found that frequent use of AI chatbots for health advice is correlated with belief in anti-vaccine myths, including false claims that vaccines cause autism or that the me…
A new study applies Bayesian decision theory to improve Large Language Models' performance in subjective tasks like tutoring and peer review, finding that Bayesian methods outperform risk-averse appro…
SONAR, a new model for multimodal embeddings, reveals that certain dimensions are highly sensitive to changes and can detect anomalies in AI understanding. By enabling consistent encoding and decoding…
Researchers introduced REAR, a training-free framework that extends test-time scaling to preference alignment for large language models. By decomposing the reward function into question-specific and u…
MemDelta, a new evaluation protocol for AI memory systems, reveals that performance varies significantly across model families, with retrieval-augmented generation (RAG) and full-context models showin…
New research reveals that diffusion large language models (dLLMs) are highly sensitive to prompt templates, with performance varying drastically based on input design. Single-template evaluations can …