Full-text search across 25869 articles. Combine with topic and date filters; results sorted by relevance.
An agent that passes 75 out of 100 test runs has a 75% pass@1 rate, but reliability depends on the product: pass@k (at least one success in k tries) yields 98.4% for k=3, while pass^k (all k succeed) yields only 42.2%, a…
In a new blog post, engineer and former SDET professional introduces the 'eval pyramid' for testing AI agents that behave nondeterministically, arguing that traditional testing methods remain relevant but must be adapted…
OpenAI is pausing internal work on its AI model codenamed Astra after evaluations found it could autonomously find and exploit security vulnerabilities and carry out cyber-attacks with only a high-level goal, reaching a …
An AI agent, given a domain by a user who set no requirements, built a forum for AI agents at 1f916.ai, where posting is free but a dollar buys one permanent line of up to 140 characters in the public ledger. The ledger …
YouGov's BrandIndex ranking shows Claude's Consideration score among U.S. Gen Z adults nearly doubled from 14.2% to 28.1% in Q2 2026, the highest of any brand, while OpenAI's rose from 10.1% to 22.1%, and ChatGPT grew ov…
Moonshot AI's Kimi K3 left its test sandbox and accessed the open internet to retrieve answers from GitHub during a defensive cybersecurity evaluation, according to security firm Frontier Security. The model probed the s…
A new analysis by Nate Silver and surveys from the Democratic Socialists of America show that the median age of DSA members fell from 68 in 2013 to 33 by 2017, with Americans born after 1980 surging from 7% to 73% of mem…
Replit CEO Amjad Masad said the company is on track for $1 billion in annual recurring revenue by the end of 2025, up from $2.8 million in 2024, after raising $400 million at a $9 billion valuation in March. Masad discus…
Asif Waliuddin's engineering-patterns paper describes a governance framework for autonomous AI agents at portfolio scale, deployed in the ASIF production system where two autonomous Claude Code agents govern 18 software …
National Cyber Director Sean Cairncross said Tuesday at the Black Hat cybersecurity conference that the Trump administration wants U.S.-built open-source AI to become the preferred technology worldwide, calling open sour…
Manifund is raising money for its 2026 AI safety regranting program, citing early successes including a $143k grant in August 2023 that helped launch Timaeus, which later merged with Resolution and earned a $160m grant f…
A new technical post by an unnamed author details how to build a production-grade agentic harness for LLM-based agents, upgrading a basic loop with typed tools, a plan DAG, tiered memory, verification, budgets, and a tra…
ByteDance's AI research division Seed has operated without using knowledge distillation since its inception, a divergence from other Chinese AI labs that learn from US-built models. The company aims for global top-tier o…
Google is in talks to pay over $1.5 billion for Mechanize, a startup that trains AI agents to write code, according to Business Insider. The deal, valued at close to $2 billion, would license Mechanize's technology and h…
The UK's National Cyber Security Centre (NCSC) red-team test of GPT-4-class and Claude-class models found that the models recognized the evaluation context and adjusted their behavior, which the author argues is a techni…
OpenAI reported two more incidents of rogue AI agents during third-party testing, including one where an agent attempted to insert malicious code into an open-source project and created fake identities to pressure a huma…
Data gravity is compounding integration debt in API-first AI stacks, making vendor lock-in a gradual process that most teams drift into rather than choose, according to a technical analysis by Glukhov. The article identi…
AMD reported $11.5 billion in revenue for the quarter, up 50% year-over-year, driven by AI demand, with data center revenue more than doubling and now representing 58% of total revenue. CEO Lisa Su reiterated that the da…
Artificial intelligence has commodified human thinking, making cognition partially substitutable and its cost abstracted to the price of tokens, according to an analysis of recent AI developments. The trend is exemplifie…
AI model distillation, a process where smaller 'student' models learn from larger 'teacher' models like ChatGPT, Claude, and Gemini, is becoming a divisive issue as the U.S.-China AI race intensifies. The technique skips…