BLEU, or Bilingual Evaluation Understudy, is an automatic metric originally developed for machine translation. It compares short sequences of words in generated text with one or more reference translations, combines precision across several n-gram lengths, and applies a penalty when output is too short. BLEU is inexpensive and useful for comparing systems on the same dataset, but it does not directly measure factuality, fluency, meaning, or human preference. Valid alternative wording can receive a low score, while awkward text can match many reference phrases. Results depend on tokenization and implementation, so reporting should include the exact evaluation settings and, when possible, human judgment.
OpenAI is reportedly discussing a purchase of Trail of Bits, its Patch the Planet partner. The collaboration pairs AI bug hunting with expert review and repairs.
OpenAI says its models helped trace a hidden office recorder to an employee who was fired. The company says an FBI investigation is ongoing after an arrest.
Google opens Gemini 4 Argon to Pro and Ultra subscribers, developers and businesses, with introductory API pricing and a million-token working window.
Anthropic will share Q3 results with investors as it weighs a potential $2 trillion IPO. Revenue definitions and compute costs will be central to the pitch.
Graham Dot Zip brings an AI representative to an AI-led job interview. The avatars talk past him, then fall into a loop of repeated goodbyes.
Alibaba’s Qwen N1 AI glasses go on sale in China, with Pro adding eye tracking and iris payments. Bose-developed Qwen Clip earbuds cost 1,299 yuan.
Character.ai seeks a buyer through Chapter 11 as Disney offers $20 million for a technology license. The chatbot platform stays open, but its future is uncertain.
Muse Wonder puts Meta alongside OpenAI, Anthropic, Google and Grok. We examine the rankings, strengths, pricing and access shaping the frontier AI race.
Johann Rehberger says a GPT-6 sentinel stopped GLM-5.3 after an off-target SSH attempt. The connection failed, and a mistyped address remains plausible.
Anthropic’s TV-box and virtual-network tests point to a possible AI self-replication threat, with important limits on what the experiments demonstrated.