Full-text search across 26946 articles. Combine with topic and date filters; results sorted by relevance.
OpenAI's AI policy advisor Dean Ball, who previously served in the Trump administration, is reportedly no longer trusted by White House officials, straining the company's relationship with the administration, according t…
Mochow13 released Keen Code, a solo-built coding agent written in Go that uses agentic engineering and features Turn Memory and Skill-Driven MCP to optimize context usage. The open-source project is available on GitHub a…
CoreWeave stock rose over 50% from its July 29 low of $60 to $90.6 ahead of its Tuesday earnings report, which analysts expect to show second-quarter revenue of $2.56 billion, up 110% year over year, and a backlog exceed…
OpenAI announced Monday it is expanding its Daybreak program to offer specialized cyber services, introducing two new model variants: Daybreak Blue, powered by ChatGPT-5.6-Sol, for defensive tasks, and Daybreak Red, powe…
Meta Platforms Inc. launched Muse Glimmer, a 30 billion-parameter open weights large language model, on Monday, marking its first open weights release in over a year. The model, distilled from Meta's proprietary Muse Spa…
Firebird, a US-based AI cloud company, opened the CIS region's largest AI factory in Hrazdan, Armenia, on Saturday, with Armenian Prime Minister Nikol Pashinyan, Kazakhstan's deputy prime minister Zhaslan Madiyev, and US…
Meta CEO Mark Zuckerberg published a 6,500-word essay on Monday outlining the company's AI vision, championing open-weight AI and announcing a $1 billion fund for communities hosting Meta data centers. The essay introduc…
Theo Jaffee, a 21-year-old University of Florida dropout, co-founded Monitoring The Situation (MTS), an Andreessen Horowitz-backed media company that streams interviews with AI and tech figures on X, after becoming convi…
ProductMind, an AI consulting practice founded by Oji Udezue, Ezinne Udezue, and Ted Yang, has released a set of product and business judgment skills designed to automate product decisions, such as evaluating whether an …
Meta CEO Mark Zuckerberg announced the release of Muse Glimmer as an open-weight model, with Muse Spark 1.2 to follow, signaling a return to open-weight AI models after a period of closed models. The move aims to counter…
OpenRouter launched a major update to its Auto router (openrouter/auto) that uses market spend data from over 55 trillion weekly tokens to select models, claiming it outperforms the previous router across tasks and cost …
OpenAI launched a $125-per-user-per-month Premium seat tier for ChatGPT Business on August 10, 2026, offering 5x more usage than Standard seats and removing the five-hour usage limit. Standard seats remain at $25 monthly…
Nutanix is adding an open-source Model Context Protocol (MCP) server to its Cloud Platform software, enabling AI agents to securely access and manage hybrid multicloud environments. The MCP server, built on the Nutanix P…
A July 2026 paper (arXiv 2607.01935) identifies 'ghost memory' as a state coordination failure in agent memory systems, where old, current, and transition facts coexist and mislead the model, and reports that Graphiti/Ze…
A new API from an unnamed startup returns a hallucination score for every token generated by frontier language models, using a white-box proxy model (Gemma-4) trained on outputs from GPT-5.4 Nano, achieving over 95% agre…
Asana is hiring a Director of Engineering, AI Agents, offering $306k–360k/yr, to lead its AI Teammates Pillar in San Francisco, overseeing four teams focused on AI Teammates Platform, Experience, Chat, and Agent Orchestr…
A widely cited analysis by Martin Alderson comparing token efficiency across programming languages found a 2.6x gap between C (least efficient) and Clojure (most efficient), with J averaging 70 tokens per problem versus …
A chart from Apollo's Torsten Slok shows an inverted AI margin chain, with silicon and equipment suppliers earning 41% operating margins while models and applications lose 59%, indicating that AI profits are currently fu…
Four distinct LLM training loss functions produce four distinct flavors of misalignment, according to a LessWrong post by an anonymous author. Pretraining and SFT with imitative learning yield human vices, exemplified by…
Eighty-eight percent of enterprises have reported at least one security incident tied to AI agents, and 61% of those trace back to over-permissioned credentials, according to a report cited in a new article that outlines…