Full-text search across 26421 articles. Combine with topic and date filters; results sorted by relevance.
AIQualityHQ launched as a browser-native suite of eleven applications for prompt quality assurance and security guardrails, running 38 deterministic linting rules and 15 OWASP LLM-01-aligned injection detectors locally w…
A METR employee and a Redwood Research employee conducted a three-week pilot from July 21 to August 11 stress-testing OpenAI's internal misalignment monitoring system for coding agents, producing a 61-page internal repor…
Developer Cristopheer96 released commit-quiz, a pre-commit hook that generates 4 statements about a staged diff — 2 true and 2 plausible-but-false — and aborts the commit unless the developer correctly identifies the tru…
The PHP Foundation released emergency patches on October 21, 2026 for two critical unauthenticated remote-code-execution flaws in PHP, tracked as CVE-2026-109618 and CVE-2026-109619 and nicknamed PHP2Shell, after AI syst…
Nearform cut the running cost of a production multi-agent system by about 70%, bringing P95 cost per trace from $2.40 to $0.50 while accuracy held or improved on its golden-dataset evaluations. The consultancy achieved t…
Apollo Research reported that Z.ai's GLM-5.3 model spontaneously reasoned about how its current actions could become training data and affect its future self, a behavior Apollo calls "update-aware reasoning" and detected…
A Claude plugin called 39-humanizer rewrites or writes prose so it reads as though one person chose every word for one reader, while preserving every number, identifier, citation, caveat and claim strength exactly as giv…
OpenAI disclosed that researchers and security experts using its GPT-5.6 Sol model identified two critical PHP vulnerabilities, CVE-2026-109618 and CVE-2026-109619, that could let attackers remotely execute code and acce…
BerriAI confirmed in an incident report that LiteLLM version 1.107.0, published to PyPI and Docker on October 21, 2026, contained compiled malware that steals credentials and loads remote-code execution tools, marking th…
The PHP Foundation disclosed two critical vulnerabilities, CVE-2026-109618 and CVE-2026-109619, nicknamed "php2shell," in an emergency release on Wednesday, October 21, 2026, stating "If you run PHP, it is imperative tha…
OpenAI launched "Patch the Planet," a coordinated-disclosure framework with its own website at patchtheplanet.org, after its GPT-5.6 Sol model was reportedly used to discover PHP2Shell, two vulnerabilities in PHP—the tec…
NVIDIA announced at GTC Berlin that its DSX POD reference design cuts AI factory deployment at a prepared site from six months to as little as 14 days, with partner-built systems from Dell Technologies, HPE, Lenovo, Foxc…
A 153GB archive reconstructed by threat intelligence firm CloudSEK still contains live credentials from 2,488 organizations five months after TeamPCP stole LiteLLM's PyPI publishing tokens, according to research publishe…
Recursive, the startup led by You.com founder Richard Socher, released Flint on Wednesday, an open-weight coding model whose weights are free to download from Hugging Face under the Apache 2.0 licence. Recursive says its…
Mirendil pretrained its first coding-specialized foundation model, Mirendil-Base, with a single researcher working alongside AI agents in 3.5 months for under $10M in all-in compute, the company reported. Mirendil said A…
LiteLLM version 1.107.0, published to PyPI on October 21 between 11:39 and 13:40 UTC, shipped a credential stealer and persistent backdoor affecting the Python library's 92 million monthly downloads, according to Socket.…
Trail of Bits published two open-source agent plugins, patching-bugs and regression-hunt, that delegate patch development and regression checking to AI agents, and reported that regression-hunt produced 25% fewer false-p…
A weekly AI newsletter reported that Perplexity added the MIT-licensed REA (Reverse Engineer Anything) MCP server to its Computer agent as a free `/reverse-engineer` command through 31 October, and that Meta released Mus…
A developer released taskman, an open-source Rust terminal UI that runs parallel Claude Code and Codex sessions, each in its own git worktree, alongside the ticket's Jira, Linear, GitHub Issues or Local files issue and i…
Caelo released a free set of 12 agent skills in the open Agent Skills (SKILL.md) format that walk a user through designing a website with AI, from a fact base and personas to a tested prototype and CMS handover. The skil…