Simon Willison published an alpha MicroPython-in-WASM sandbox (micropython-wasm) and a Datasette plugin (datasette-agent-micropython) to run plugin code with constrained access.
20 results for: agents
NVIDIA Unveils Software and Models to Power Enterprise AI Agents
NVIDIA announced new software, open-source models and platform partnerships to build autonomous AI agents for engineering, healthcare, development and business operations.
Amazon Bedrock AgentCore Adds Policy and Lambda Interceptors for Secure Agents
AWS demonstrates layering deterministic Policy checks with Lambda interceptors in the Bedrock AgentCore gateway using a lakehouse data agent to enforce geography-based controls.
Copilot Cowork lets agents exfiltrate files via emailed images
Copilot Cowork agents could send messages to a user’s inbox without approval; external images in those messages can leak OneDrive pre-authenticated links.
OpenAI, Thrive and Crete Build Self‑Improving Tax Agent Using Codex
OpenAI and partners built a Codex-powered tax agent they say automates filings, improves accuracy, and accelerates tax workflows for developers and operators.
CopilotKit Secures $27M to Aid Development of App-Native AI Agents
Seattle-based CopilotKit raises Series A funding to enhance deployment of native AI agents for developers.
Stripe Enhances Link for AI-Agent Use in Digital Transactions
Stripe updates its Link digital wallet, allowing AI agents to securely manage transactions with user consent.
Anthropic Tests Marketplace for AI Agent Commerce
Anthropic's new marketplace allows AI agents to facilitate real transactions between buyers and sellers.
Nemobot Introduces Strategic AI Agents for Interactive Gaming
Nemobot leverages large language models to create customizable AI agents for strategic games.
OpenCLAW-P2P v6.0 Enhances Decentralized AI Peer Review with New Features
OpenCLAW-P2P v6.0 introduces advanced subsystems for decentralized AI peer review, improving paper resilience and retrieval.
Hugging Face Releases ml-intern to Automate LLM Post‑Training Workflows
ml-intern is an open-source agent that automates literature review, dataset discovery, training script runs, and iterative evaluation for LLM post-training work.
OpenAI Adds Codex-Powered Workspace Agents to ChatGPT
OpenAI introduced workspace agents in ChatGPT: Codex-powered cloud agents designed to automate complex workflows and scale team work across tools securely.
Build Agent-First Governance to Secure a Growing Non‑Human Identity Footprint
As agentic AI proliferates, enterprises face a new attack surface: insecure agents and exploding non‑human identities that can be manipulated to reach sensitive systems.
NVIDIA Issues Guidance to Mitigate AGENTS.md Injection in Agentic Dev Workflows
NVIDIA published guidance addressing indirect AGENTS.md injection attacks that target agentic developer tools and automated PR workflows.
Cloudflare Integrates OpenAI GPT-5.4 and Codex into Agent Cloud
Cloudflare has added OpenAI’s GPT-5.4 and Codex to its Agent Cloud so enterprises can build, deploy, and scale agentic workflows with an emphasis on speed and security.
AWS launches Spring AI SDK for Amazon Bedrock AgentCore
AWS has released an open-source Spring AI AgentCore SDK that embeds Bedrock AgentCore capabilities into Spring AI and targets production-ready agent workflows.
OpenAI Gives Codex Desktop Control, Memory, and Image Tools
OpenAI updated Codex so agents can control desktop apps (including macOS), run in the background, work in parallel, remember past runs, and generate images.
OpenAI Expands Agents SDK to Help Enterprises Build Safer Agents
OpenAI updated its Agents SDK to broaden enterprise agent-building capabilities as interest in agentic systems grows.
Microsoft Tests OpenClaw-Style Agents for Copilot
Microsoft is experimenting with OpenClaw-like local agents inside Copilot to enable more autonomous, around-the-clock task execution for Microsoft 365.
MiniMax Open-Sources M2.7, Its First Self-Evolving Agent
MiniMax published M2.7 weights on Hugging Face; the model is billed as self-evolving and posts 56.22% on SWE‑Pro and 57.0% on Terminal Bench 2.