
AWS adds persistent instances for AI agents
Bedrock AgentCore will let agents run for several days, share sessions, and use GPUs when a task requires them.
Read publication →
Bedrock AgentCore will let agents run for several days, share sessions, and use GPUs when a task requires them.
Read publication →
Anthropic used Claude as an automated researcher to find methods addressing ten types of unwanted behavior in AI models.
Read publication →
Future Claude models will include a statistical signal to estimate whether the AI participated in a text without identifying its author.
Read publication →
Cloudflare introduced tools to check whether agents can read a website and whether AI assistants cite or mention it.
Read publication →
Kitesurf lets agents open and process web pages without running every feature included in a traditional browser.
Read publication →
The proposal will let users assign funds and limits so agents can pay for APIs, tools, and digital content.
Read publication →
The experimental model can describe images, read screenshots, and analyze charts through the DeepSeek API.
Read publication →
AWS now lets applications store and query vectors directly in DynamoDB for semantic search, RAG systems, and AI agent memory.
Read publication →
Google introduced new Gemini Live features for managing email, checking schedules, and delegating tasks through spoken instructions.
Read publication →
Google introduced two speech-to-text models: one processes recordings and the other produces live transcriptions.
Read publication →
GitHub will let teams start shared Copilot sessions from Slack and Microsoft Teams to investigate issues and prepare code changes.
Read publication →
The new model can generate, extend, and edit videos through instructions, with transitions between images and output up to 4K through upscaling.
Read publication →
IBM will integrate OpenAI models and tools into its consulting services to modernize business processes, applications, and security.
Read publication →
Jetson Orin Nano 2 doubles inference performance over its predecessor and improves energy efficiency, according to NVIDIA.
Read publication →
The new CPU is designed to handle tools, searches, and coordination tasks used by artificial intelligence agents.
Read publication →
OpenAI offers Zero Data Retention to eligible API customers and is preparing safety controls for longer-running tasks.
Read publication →