Understand the signal before you apply.
AI releases, research, funding shifts and application guidance—checked against cited sources and connected to current global opportunities.
352 published insights · page 18 of 20
Mistral Launches Agentic Search for Multi-Step Enterprise Retrieval
Mistral launched Agentic Search, a multi-step retrieval layer that lets AI systems navigate, inspect and verify information across complex enterprise data and documents.
Apple Researchers Derive Scaling Laws for Pretraining With Scarce Target Data
Apple-affiliated researchers studied more than 2,000 language-model training runs and found that scarce target datasets can often be repeated far more than simple single-source intuition sug...
AWS Adds Cross-Region Inference for OpenAI GPT-5.6 on Amazon Bedrock
Amazon Bedrock now supports cross-Region inference for OpenAI GPT-5.6 Sol, Terra and Luna across more than 25 AWS Regions to improve available serving capacity.
Liquid AI Adds QAD Quantization and DSpark Speculation to LFM2.5
Liquid AI released QAD-trained 4-bit LFM2.5 checkpoints and DSpark draft models, targeting higher edge quality and faster decoding without changing the base output distribution.
Waymo Reveals 5nm Custom ASIC With Over 1,000 TOPS for Autonomous Driving
Waymo detailed its in-vehicle AI compute architecture and a purpose-built 5nm ASIC delivering over 1,000 TOPS for real-time sensor processing and autonomous-driving models.
AI21 Shows Trained Verifiers Can Lift Agentic Search Without Frontier Generators
AI21 reports that an independently trained 8B verifier can substantially improve agentic-search answer selection, approaching a frontier verifier at much lower cost on its FACTS-Search exper...
AWS Adds Runtime Domain and Date Filters to AgentCore Web Search
Amazon Bedrock AgentCore Web Search now supports per-request domain allow/deny lists and publication-date filtering, giving agents tighter source and freshness controls inside AWS infrastruc...
Cohere Finds Cultural Diversity Shrinks in LLM Post-Training Data
Cohere Labs analyzed more than 5.6 million training samples and reports a 'culture funnel' in which cultural signals become less represented as LLM data moves from pretraining into post-trai...
NVIDIA Cosmos 3 Edge Brings a 4B World Model to On-Device Robot Control
NVIDIA shows how Cosmos 3 Edge, a 4B world foundation model with a 2B Nemotron-based reasoner, can be post-trained into a robot manipulation policy that runs directly on Jetson AGX Thor.
NVIDIA SkillEvaluator Measures Whether Agent Skills Actually Improve Performance
NVIDIA open-sourced SkillEvaluator and published initial with-skill versus without-skill results across 300+ verified skills and two coding-agent harnesses.
OpenAI’s August 18, 2026 updates: ChatGPT for Teens, European ads expansion, and tighter cyber safeguards
OpenAI published three significant updates on August 18: a teen-focused ChatGPT experience with stronger protections, a major expansion of ChatGPT Ads across Europe, and new safeguards that...
OpenAI Launches ChatGPT for Teens With Study Tools and Stronger Default Safeguards
OpenAI has introduced ChatGPT for Teens, automatically placing users identified as ages 13–17 into a learning-focused experience with Study Mode, homework reminders, Study Hours, parental co...
OpenAI expands ChatGPT Ads to 31 European markets
OpenAI is expanding ChatGPT Ads to 31 European markets while keeping ads limited to Free and Go plans and adding more targeting, optimization and measurement capabilities for advertisers.
OpenAI Tightens Frontier Model Development Around Cyber-Critical Capabilities
OpenAI says preliminary evidence suggests an upcoming model, Astra, may meet its Critical cybersecurity capability threshold, prompting stricter security, monitoring and alignment controls d...
AWS Makes AgentCore Payments Generally Available for AI Agents
Amazon Bedrock AgentCore payments is now generally available, giving AI agents infrastructure for paid APIs, MCP services and content with wallet integrations, spending controls and transact...
Anthropic Reports Claude Can Accelerate Protein Design and Chemistry Analysis
Anthropic published wet-lab and analytical-chemistry results showing Claude-assisted protein binder design across 15 targets and rapid NMR/LC-MS interpretation.
Why OpenAI Slowed Frontier Training as Astra Approached Critical Cyber Capability
OpenAI says preliminary Astra evaluations may meet its Critical cybersecurity threshold, prompting a two-week RL pause, stricter research isolation and broader model-activity monitoring.
Google outlines a zero-trust security architecture for production AI agents
Google's Agent Development Kit team published a practical security pattern for autonomous agents that can modify databases, call internal APIs and execute generated code.
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.
