Full-text search across 70168 articles. Combine with topic and date filters; results sorted by relevance.
A developer at HumanBound argues that AI agents' biggest vulnerability is untested adversarial behavior, not missing firewalls or infrastructure controls. Citing incidents like OpenClaw deleting a Meta AI Safety lead's i…
Researchers released OpenLanguageModel (OLM), an open-source PyTorch library for building and pretraining small language models with readable, composable code. The library includes 27 presets across nine model families, …
A developer eliminated 200 lines of fragile JSON-parsing code from their JobSearch tool by migrating to Anthropic's Claude tool-use API. The tool, which ingests job offers and returns structured analysis, previously requ…
Vidmoat, a new video editor built for AI agents, allows users to edit videos by giving plain-language commands to an integrated agent or connecting external agents like Claude Code and Cursor via the Model Context Protoc…
Google DeepMind's Gemini 3.5 Flash-Lite, a lightweight multimodal model described in the Gemini API documentation as a stable model supporting code execution, function calling, search grounding, structured outputs, and t…
Vercel's AI Gateway now offers Laguna S 2.1, an open-weight Mixture-of-Experts model from Poolside that supports up to 1M tokens and agentic coding tasks. The model achieves 70.2% on Terminal-Bench 2.1, 78.5% on SWE-benc…
Shadow AI—employees pasting sensitive data into ChatGPT—is rampant because it helps them work faster, and bans only drive it underground. Framz.io built a Shadow AI Data Leak Scanner that runs on its own hardware, detect…
Realbotix, the Nevada-based company known for AI-powered companion robots for intimacy, will deploy its humanoid assistant 'Sally' at Salamanca High School in western New York this September to help teach coding, robotic…
A developer released a production-shaped LangGraph multi-agent template that wires five critical production features—structured routing, parallel-branch reducers, per-node retry policies, hard budget ceilings, and resuma…
Ora has launched Ora Core, a private AI software that compiles large language models up to 70 billion parameters to run on consumer GPUs with less than 1% accuracy loss, enabling local inference without data leaving the …
Agent Native Analytics, an open-source tool built on the Agent-Native framework from Builder.io, offers a Fullstory alternative that lets teams self-host and query behavioral data with AI agents instead of fixed dashboar…
An engineer built PRForge, an AI code review agent that reviews pull requests in under 60 seconds and auto-merges them if they meet quality rules. It works across GitHub, GitLab, Bitbucket, and Azure DevOps, and includes…
Moonshot AI's Kimi K3, a 2.8 trillion-parameter open-weight AI model that sparked US tech stock declines, has paused new subscriptions due to overwhelming demand exceeding GPU capacity. The Chinese company said it is add…
Developer Cole Hellman built a personal MCP server called Codicil that indexes documentation for AI coding assistants. When the embedding service went down for two weeks, Hellman didn't notice because the system had sile…
Four research teams demonstrated in July 2026 that AI agents share a fundamental security flaw: they trust inputs without verification. Manifold Security showed that Anthropic's Claude for Chrome can be hijacked by a mal…
Adam Kues of Searchlight Cyber spent about $25 of model time using OpenAI's GPT-5.6 "Sol Ultra" to chain a routine SQL injection in WordPress (CVE-2026-60137) with a batch-route confusion bug (CVE-2026-63030) into a pre-…
Anthropic is testing a new placement for Projects in the navigation sidebar of its Claude desktop app, moving the Chat/Cowork mode toggle to the top of the window, mirroring changes OpenAI made to ChatGPT in June and Jul…
Moonshot AI's Kimi K3, a ~2.8-trillion-parameter open-weight model released in mid-July 2026, trails Anthropic's Claude Fable 5 and OpenAI's GPT-5.6 Sol overall but beats closed models on coding, agent, and engineering t…
OpenAI has reduced the context size of its Codex model from 372,000 to 272,000 tokens, according to a developer note. The change affects the model's ability to process longer code sequences in a single request.
It is too late to make technology ethical because the focus should have been on the people who build it, not on the technology itself, argues a commentator. The author contends that ethical judgment must be cultivated in…