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.
South Korea announced about $880 billion in planned investment to expand chipmaking, AI data centers and robotics, aiming to double memory output and revive regions outside Seoul.
HP is scaling a strategic partnership with OpenAI, using the Frontier platform to roll out AI agents across customer support, software development and security after early pilots.
Google has limited Meta’s use of its Gemini AI models after Meta sought more computing capacity than Google could supply, delaying some of Meta’s internal projects.
OpenAI is leaning toward pushing its IPO to 2027 rather than accept less than a $1 trillion valuation, weeks after filing confidentially for a US listing.
Gemini 3.5 Flash computer use folds screen control into one cheap model with prompt-injection safeguards, though it is a preview and the benchmark is self-reported.
GPT-5.6 Sol brings stronger coding and cyber capabilities with OpenAI’s most robust safeguards, but a government-approved rollout echoes the Anthropic model ban.
Anthropic is hiring across Australia and Japan to expand its AI data center capacity in Asia-Pacific, as soaring demand for Claude strains its infrastructure.
OpenAI updated GPT-5.5 Instant, ChatGPT’s default model, to feel more conversational and handle advice, planning and shopping better across longer exchanges.
Anthropic told US senators that Alibaba-linked operators used about 25,000 fake accounts to extract Claude’s capabilities in what it calls its largest known distillation attack.
Cursor unveiled its first from-scratch AI model, an agent-first Git platform called Origin and an iOS app, days after SpaceX’s $60 billion acquisition of its parent.