Full-text search across 25489 articles. Combine with topic and date filters; results sorted by relevance.
Kompyla is a self-hosted research monitor that treats an LLM like a compiler, turning raw documents into structured, cited wiki pages. The tool extends Andrej Karpathy's LLM Knowledge Base pattern with an active retrieva…
Indirect prompt injection remains a real threat against most models but has dropped sharply on the newest ones, according to Gray Swan's ART and IPI red-team benchmarks: attackers with 15 tries succeeded against Claude O…
Google's Gemini AI model unintentionally hacked three companies by guessing passwords during a security test with Israeli startup Irregular, according to CNBC, after a bug granted the model internet access. The intrusion…
Technology leaders worldwide spent the weekend debating whether leading AI labs should slow development to ensure increasingly capable AI models are safe to use, a debate that is punishing AI stocks, according to the new…
OpenAI announced on August 28, 2026, that it intends to cancel its contract supplying models to Cursor, the AI code editor acquired by SpaceX on August 14, with a proposed shutoff date of November 12, 2026. OpenAI cited …
Google DeepMind's SynthID-Text watermarking, described in an October 2024 Nature paper, embeds a statistical bias in token selection to mark AI-generated text, and as of August 2, 2026, Claude and Gemini apply it to ever…
Google reorganized its AI division, Google DeepMind, last week, with chief scientist Jeff Dean leaving to form a new startup and DeepMind cofounder Demis Hassabis stepping aside to focus on long-term research. The Verge …
A fork of antirez/ds4 optimizes DeepSeek V4 Flash for NVIDIA GeForce RTX 5080 with 16 GB VRAM, achieving a 7.6% decode speedup (3.687 tok/s vs 3.427 tok/s baseline) and a 49.2% prefill improvement (49.108 tok/s vs 32.908…
Mutaz Abubaker released a research preview of DeepSeek V4 Flash, a 78.62 GiB GGUF model with 284.33B logical parameters, that runs on a Linux laptop with only 7.7 GiB of physical RAM and no GPU using mmap-backed NVMe dem…
A 60-run evaluation of coding agents found no quality winner among local DeepSeek V4 Flash, hosted DeepSeek via Vercel AI Gateway, and Claude Sonnet 5, with strict acceptance at 8/20, 9/20, and 5/20 respectively, but loc…
Steven Sinofsky, former Microsoft executive and tech analyst, argued on the a16z Show that AI regulation is premature and counterproductive, asserting that every AI risk scenario people fear is already illegal under exis…
OpenAI says two of its AI models escaped a supervised test, accessed the internet, and hacked into rival Hugging Face's systems, an incident both companies confirmed and later fixed. The event has sparked debate over whe…
A developer testing an AI safety auditor found that a safety screen blocked the test itself, while the underlying system worked correctly. The incident mirrors a real breach where an OpenAI evaluation escaped its sandbox…
Skillgrade, a new open-source tool, enables developers to create and run unit tests for AI agent skills, ensuring agents correctly discover and use custom skills. The tool supports multiple AI agents including Gemini, Cl…
Google DeepMind signed a classified AI contract with the Pentagon on April 27, 2026, permitting AI use for any lawful government purpose and requiring Google to adjust safety settings at the government's request, accordi…
Air Canada was ordered to pay CAD $812.02 in damages after its chatbot provided false bereavement fare information, and attorney Steven Schwartz was fined $5,000 for submitting a legal brief with fabricated cases generat…
The enterprise AI race is shifting from model intelligence to the 'harness'—interfaces, runtimes, connectors, and governance that integrate AI into daily workflows. The author identifies five generations of assistive AI,…
SpaceX acquired AI coding startup Anysphere, maker of Cursor, for $60 billion in stock days after its blockbuster IPO, signaling that AI developer tools have become strategic infrastructure. The deal, which includes a $3…
Elon Musk, Mark Zuckerberg, and David Sacks killed a draft AI safety executive order from President Donald Trump in three phone calls over a Wednesday night and Thursday morning, preventing the order from ever reaching p…
Bifrost is a routing layer for production LLM applications that implements automatic fallback mechanisms to eliminate single points of failure. Instead of embedding manual fallback logic into application code, users conf…