Full-text search across 26942 articles. Combine with topic and date filters; results sorted by relevance.
Sending every LLM request twice and taking the faster response outperforms paid priority tiers, cutting worst-case time to complete response from 9.8s to 3.5s and worst-case time to first token from 4.2s to 1.2s in tests…
At an OpenAI summit on the future of mathematics, mathematician and Fields Medalist Timothy Gowers warned that AI could lead to a stagnation in mathematical progress despite superhuman AI capabilities. He cited a surge i…
Prime Intellect has open-sourced Prime Agent, a framework for building self-improving recursive language model (RLM) agents that operate in a persistent Python REPL. The framework enables agents to retain state, spawn su…
Demis Hassabis stepped down as head of DeepMind, and Jeff Dean, Sanjay Ghemawat, Oriol Vinyals, and Quoc Le—three of the four leads of Gemini—left Google to found a startup focused on recursive self-improvement (RSI), ac…
OpenAI has lost its head of ethics, Chloé Bakalar, who left in July less than a year after joining, with no replacement lined up, according to the Financial Times. The departure comes as OpenAI faces scrutiny over its AI…
Nvidia unveiled NeMo Switchyard, a software router that directs AI prompts to different models to cut costs, claiming a 74 percent reduction in job completion costs versus using Claude Opus 4.8 alone, with a six-point ac…
A developer advises moving marketplace review summarization, tagging, and extraction to batch LLM jobs when no customer is waiting, while keeping realtime calls for interactive work. They emphasize tenant-scoped batches …
Alibaba's Qwen team launched Qwen3.8-Max on August 3, a 2.4 trillion parameter sparse Mixture-of-Experts model that runs at over 4,000 tokens per second per GPU on Nvidia's GB300 NVL72 hardware, scoring 86.1 on OSWorld-V…
The National Institute of Standards and Technology (NIST) published a Request for Information on Wednesday seeking input on modernizing its National Vulnerability Database (NVD) with artificial intelligence to enhance ho…
A developer explains that truncated AI responses are not mysterious because APIs provide stop reasons, yet most integrations ignore them. The post recommends adding a boundary assertion to check finish_reason or stop_rea…
Meta Superintelligence Labs released Muse Glimmer, a 30-billion-parameter dense open-agentic model under Apache 2.0, designed for local agents with multi-step reasoning, tool calls, image understanding, and failure recov…
GamesBeat Next will return to San Francisco on November 2-3, 2026, bringing together CEOs, founders, CMOs, investors, and business leaders from gaming, AI, entertainment, media, sports, and technology to examine trends i…
Israeli cybersecurity firm A Security reported that a researcher used fewer than 20 AI prompts to find and exploit vulnerabilities in Zoom's annotation tool, building a working attack in under 24 hours that could let an …
In August 2026, AI-related content dominated every post on the blog, prompting the author to plan a return to non-AI topics such as childhood, education, fertility, housing, and dating. The roundup also highlights widesp…
Red Planet Labs reports that large language models (LLMs) are now consistently one-shotting complex backend challenges, including an unbalanced social graph, a social graph with fanout, and a Slack-like chat backend, as …
Researchers at Israeli cyber firm Dream reported the first publicly known 'near-autonomous' AI attack on a government target, in which suspected Chinese hackers used open-source AI models to steal more than 2,500 personn…
Arkency has built Planet Arkency, a multi-tenant knowledge graph with a closed ontology, using Rails Event Store and an LLM to maintain organizational knowledge from unstructured sources like meeting transcripts and Slac…
PortSwigger researcher James Kettle unveiled the HTTP Terminator, an autonomous AI system that invented novel HTTP desync attack techniques and compromised banks, security solutions, and government infrastructure, at Bla…
OJCP (Open Job Consumption Protocol) aims to fix the chaos of AI agents applying for jobs by standardizing job searches and applications through MCP tools, a manifest at `/.well-known/ojcp.json`, and schemas extending sc…
Perplexity has begun blocking ads served to AI agents on Time's website, warning publishers that running such 'deceptive advertising' could result in downranking in its search index. The ads, created by adtech company Mo…