Understand the signal before you apply.
AI releases, research, funding shifts and application guidance—checked against cited sources and connected to current global opportunities.
213 published insights · page 5 of 12
Anthropic Says Claude Helped Surface 2,300 Disclosed Open-Source Vulnerabilities
Anthropic’s coordinated disclosure dashboard says Claude models helped identify vulnerabilities that led to 2,300 disclosures across 392 open-source projects, with human triage and maintaine...
StreamPI Adds Efficient Temporal Memory to Vision-Language-Action Robots
HKU and ACE Robotics researchers introduce StreamPI, a temporal modeling method that gives a pretrained single-frame VLA multi-frame memory without adding model parameters.
EVOMAL Finds Self-Poisoning Risk in Self-Evolving Coding Agents
A new arXiv paper identifies a self-poisoning attack in coding agents that write and reuse their own skills, showing malicious patterns can propagate through shared skill libraries even afte...
AIRSEAI Joins LF AI & Data to Build Open Embodied AI Infrastructure
LF AI & Data Foundation is now hosting AIRSEAI, an open-source embodied-AI software stack intended to reduce robotics fragmentation by supporting reusable intelligence across different hardw...
Google Cloud Brings Long-Context Multimodal Embeddings to TPU via vLLM
Google Cloud added native TPU support in vLLM for long-context text and multimodal embedding serving, with GKE elasticity and open deployment recipes.
AWS and NVIDIA Plan 2 Million Additional GPUs for 2027–2028 AI Expansion
AWS and NVIDIA have announced plans to deploy 2 million additional Blackwell Ultra, Rubin and Rubin Ultra GPUs across AWS in 2027–2028, alongside Vera CPUs, NVHBM integration, federal AI fac...
Sentence Transformers 6 Expands Trainable Multi-Vector Retrieval
Sentence Transformers v6 adds MultiVectorEncoder for ColBERT-style late interaction, with a new end-to-end training and finetuning workflow.
MIT CrysVCD Constrains AI Materials Generation for Better Stability
MIT's CrysVCD adds chemical valence constraints before crystal generation, improving computational stability rates while targeting useful material properties.
Broadcom AAI v26 Adds Conversational Insights and Cost Attribution
Broadcom AAI v26 is generally available with natural-language workload exploration, automation cost attribution, OpenShift containers and AI-assisted security scanning.
AWS and NVIDIA Plan 2 Million More GPUs for Agentic and Physical AI
AWS and NVIDIA announced plans to deploy 2 million additional NVIDIA GPUs across AWS infrastructure in 2027–2028 while expanding integration across CPUs, networking, open models, accelerated...
TrendAI AESIR Reaches 97% on CyberGym AI Security Benchmark
TrendAI reports that its AESIR agentic exploit-remediation engine reached 97% on CyberGym, a real-world vulnerability-analysis benchmark, highlighting the value of multi-model orchestration,...
GitHub Copilot Code Review Enters Public Preview for Azure Repos
Microsoft says GitHub Copilot Code Review for Azure Repos is now in public preview, adding automatic reviews, custom instructions, managed-pool support and project-level cost visibility.
NVIDIA COMPASS Uses AI Agents to Build Cross-Embodiment Robot Navigation Policies
NVIDIA’s COMPASS workflow combines a pretrained X-Mobility navigation policy, residual reinforcement learning and coding-agent skills with explicit human approval gates for adapting navigati...
Claude Cowork Gets a Built-In Browser for Agentic Web Tasks
Anthropic has added a separate built-in browser to Claude Cowork so the desktop agent can navigate websites, read pages, click, type and fill forms without requiring a browser extension.
Lightwheel Opens 10,000 Hours of EgoSuite Physical AI Data
Lightwheel and Hugging Face have released the first 10,000 hours of EgoSuite-Open100K, an egocentric human-activity dataset intended for physical AI and robot learning, with 90,000 more hour...
Google Research Details GlucoFM, a Dual-Stream Foundation Model for Continuous Glucose Monitoring
Google Research highlighted GlucoFM, a lightweight dual-stream foundation model for continuous glucose monitoring that separates slow glycemic trends from short-term deviations and transfers...
MLPerf Adds End-to-End RAG Benchmark for Multi-Model AI Systems
MLCommons has introduced MLPerf's first end-to-end retrieval-augmented generation benchmark, measuring both vector-database ingestion and iterative multi-hop question answering across a mult...
Llama-Mobile Compresses an 11B Vision Model to 2.7-Bit Weights for Arm CPUs
Graphcore Research and Arm's Llama-Mobile work compresses Llama 3.2 Vision 11B from 21.3 GB of bfloat16 weights to 3.7 GB using S3D8 2.7-bit storage and quantization-aware distillation for m...
Turn intelligence into applications.
Create a free alert for the topics, roles or countries that matter. We email only new verified matches.
More ways to save
Discover deals, coupons and free courses on our sister site.
