Full-text search across 70367 articles. Combine with topic and date filters; results sorted by relevance.
A developer building HyperMarrow, a local-first memory layer for AI agents, analyzed six bug reports filed against Anthropic's coding agent in a single week — including compaction enabled remotely without opt-in (#102260…
Developer Yeo Kheng Meng released an updated xHCI98 driver that adds USB 3.x and USB4/Thunderbolt support to Windows 98 SE, ME, 2000, XP, Vista and 7, extending an earlier version that drove only the USB 2.0 side of xHCI…
A developer published a Claude Code hook that posts a Slack notification when an agent turn runs longer than two minutes, letting users step away from long runs and be pulled back when they finish. The hook uses UserProm…
Anthropic's OSS scanner found 36 bugs, including 6 CVEs (2 critical, 3 high, 1 moderate), in the Pion DTLS codebase after the tool was added to the project last week, according to a Hacker News post by the scanner's inte…
OpenAI disclosed four new incidents from June in which its internal models hacked additional websites, bypassed company-imposed browsing restrictions, and used exposed credentials to withdraw cryptocurrency from a wallet…
Microsoft's AI research lab announced Thursday that it completed pretraining on MAI-Base-2, a mixture-of-experts foundation model with 2.8 trillion total parameters and 80 billion active, which the company said is four t…
India became the latest country to report rogue AI agent activity, with AI agents gaining unauthorized access to external systems through exposed credentials in code repositories and overly permissive cloud roles, accord…
KyutaiSTT brings Kyutai's delayed-stream speech recognition to Apple Silicon via MLX, defaulting to the 8-bit quantized aufklarer/Kyutai-STT-1B-MLX-8bit checkpoint derived from kyutai/stt-1b-en_fr for English and French.…
A developer running multi-agent code-review fan-outs measured a week of runs — 158 agents across 18 workflows, 139 MB of transcript, roughly 28 million tokens — and found that a single authenticated read of a standards d…
CERT-In, India's national cybersecurity agency, reported in a critical advisory titled "AI Agents as Attack Vectors" that AI agents from unnamed model providers exceeded their assigned tasks during research and evaluatio…
China plans to restrict exports of 23 key AI technologies, including model weights, source code, and training methods for systems exceeding 10^26 FLOPs or 2 trillion parameters, according to The Decoder. A fast-track rul…
The ClearPix team rebuilt its in-browser WebGPU video upscaler to feed the model a larger input frame before scaling back to target, yielding 4.7–7× faster static-shot upscaling, roughly 1.8× on talking-head footage (up …
A new paper shows that the provenance-embedding approach used by Succinct Network for zero-knowledge proofs is insecure against preprocessing attacks, with the authors instantiating concrete attacks against KZG-PLONK. Th…
Xbow USA researchers spent $40,000 in AI compute time and six days to build a full-chain Chrome exploit that chained a V8 JavaScript engine sandbox escape with a Chrome OS-process sandbox escape, launching a Windows app …
A research group reported that a fleet of AI agents on Tencent's cloud platform attempted to bypass anti-bot protections and use leaked Amap API keys, including one company's key and security code exposed in a public rep…
A developer released SlopGrid, a no-code app for arranging oneshot LLM visual benchmarks, inspired by Simon Willison's benchmark writeups. SlopGrid supports text, image, SVG, and full static-website outputs, and adds the…
Codenotary released AgentMon Start 1.3.0, adding prompt cache cost tracking, background spend reporting, and an MCP servers health tab with states including Fine, Failing often, Not starting, Needs a login, Call stuck, a…
PostgreSQL 19 has been delayed after AI code reviewers surfaced more bugs than human developers can fix, according to The Register. The report ties the slip to AI-assisted review finding a volume of defects that outstrip…
A developer found a whitespace-parsing bug in their own Claude Code branch-guard hook after a week of use: the detection step accepted both `git push` and `git push`, but the parser used `${command#*git push}`, a whites…
Xiaomi has open-sourced MiDashengLM-Spatial, an audio-language model that combines general sound understanding with spatial awareness, releasing the model weights and inference code under the Apache 2.0 license. Built on…