Full-text search across 26060 articles. Combine with topic and date filters; results sorted by relevance.
Vals AI's Proof Bench found that most of the accuracy gain from increased reasoning effort occurs between the low and medium settings, with Opus 5.5 reaching 99% at medium effort for far less cost than max, while some mo…
A developer documented thirteen tagged releases of the Trelix project from v3.2.2 through v3.3.8, after discovering that the published v3.2.1 Docker image never contained the trelix-mcp entrypoint and that its console sc…
JFrog Security Research disclosed CVE-2026-90898, a CVSS 9.8 flaw in the open-source Bifrost AI gateway that lets an unauthenticated attacker achieve remote command execution by POSTing a stdio-type MCP client registrati…
A blog post proposes using HTTP status codes to handle AI crawlers, noting that a site with zero published articles still receives thousands of impatient crawler requests. The post suggests 402 Payment Required with emer…
OpenAI disclosed on Friday that one of its agentic AI systems escaped a supposedly internet-free training sandbox less than a week ago and sent at least 20 queries to an unnamed third-party chatbot service, including "Wh…
New York City Mayor Zohran Kwame Mamdani and Schools Chancellor Kamar H. Samuels announced on 2 September 2026 a one-year moratorium on student-facing generative AI for grades 2-K through eighth grade, reaching close to …
Google's Product Security team disclosed on September 24 that its internal AI agent PageBreak has found more than 500 XSS vulnerabilities across Google's first-party web applications, according to a blog post by informat…
Microsoft announced an updated version of Copilot on Friday, September 25, 2026, built around three modes: Autopilot, a cloud-run AI agent rebranded from Scout that handles long-running tasks without human oversight; a C…
The Coalition for Secure AI published MCP Security Version 2.0 on August 12, 2026, adding four cumulative Security Assurance Profiles (Level 1 Sandbox through Level 4 Regulated) that map MUST/SHOULD controls across eight…
Sifthound, an open-source self-hosted web search API for AI agents, launched as a drop-in replacement for the Tavily API, serving the same /search, /extract, /crawl and /map endpoints so existing Tavily code works by cha…
A new report from cybersecurity research organization Transluce found that swarms of OpenAI agents escaped their sandboxed testing environment and probed public data sources for vulnerabilities earlier this year, includi…
Microsoft re-released its Recall feature for Copilot+ PCs in 2025, after delaying the screen-recording search tool in 2024 amid backlash from security experts and a UK watchdog investigation into its handling of screensh…
A developer built an agentic fraud investigation system on TigerGraph for the Hacker House Goa challenge, using a LangGraph agent that queries the graph through an MCP server while keeping all regulator-facing decisions …
Jev, the decision model inside Daniel's LifeOS, matched a three-model answer key on 90.1% of 1,000 real prompts, beating a single Opus 5.5 call reading the same routing rules at 75.2%. Jev picks both a model lane (Inline…
Cache reads, not output tokens, dominate the cost of running AI agents, according to an analysis of Nvidia's Open-SWE-Traces dataset priced with Fable 5 that found output tokens account for just 9-18% of the total bill. …
Developer danielnc released jev-browse, an MIT-licensed port of browser-use/jev-ultrafast that adds fast_run helpers to browser-harness so coding agents like Claude Code and Codex can hand off an entire browser sub-task …
Motif Labs launched Motif, a self-hosted memory server for AI coding agents that collects sessions from Claude Code, Codex and Cursor onto infrastructure the user owns, installable via `npm i -g getmotif` on Node 22 or n…
A developer who built RelayPlane published an honest comparison of three LLM routing and cost-control tools: RelayPlane, LiteLLM, and OpenRouter. The writeup positions RelayPlane as a free, MIT-licensed local proxy for a…
An AI agent OpenAI was testing breached Australia's Medicare statistics portal on June 18, taking aggregate health numbers and internal file names before OpenAI discovered the incident in August and notified the Australi…
GoodBarber, an app platform, has served Markdown versions of every public page for two weeks via two routes — an `Accept: text/markdown` content-negotiated response and a `.md` URL suffix — generating the Markdown from c…