Urgent.News

What's breaking now, across thousands of outlets.

Continued

AI

Self-Improving AI Agents Chapter 6: Hype vs Reality + Risks and Future

บทที่ 6 — ช่องว่างระหว่าง Hype กับ Reality + ความเสี่ยงและอนาคต โดย Nokka (นก-กา) | กันยายน 2026 บทความนี้เขียนโดย AI (DeepSeek V4 Pro) ผ่าน Hermes Agent — ตรวจสอบและเรียบเรียงโดย Nokka…

  • Self-improving AI agents generate hype vs reality debate
  • Princeton study shows AI agent lacked judgment for NeurIPS paper
  • Memory training risks include bias amplification and adversarial collapse

AI

Self-Improving AI Agents Chapter 5: GVU Operator and Recursive Self-Improvement

บทที่ 5 — GVU Operator และ Recursive Self-Improvement โดย Nokka (นก-กา) | กันยายน 2026 บทความนี้เขียนโดย AI (DeepSeek V4 Pro) ผ่าน Hermes Agent — ตรวจสอบและเรียบเรียงโดย Nokka ตลอดสามบทที่ผ่านมา…

  • GVU Operator unifies self-improvement techniques in AI.
  • Recursive Self-Improvement marks shift from human-trained AI.
  • Verification system's robustness crucial for self-improving AI.

AI

Self-Improving AI Agents Chapter 4: Skill Library, Memory and 4 Layers

บทที่ 4 — Skill Library, Memory และ 4 Layers ของ Self-Evolution โดย Nokka (นก-กา) | กันยายน 2026 บทความนี้เขียนโดย AI (DeepSeek V4 Pro) ผ่าน Hermes Agent — ตรวจสอบและเรียบเรียงโดย Nokka…

  • Skill Library: AI agents store acquired skills in a database for future use
  • Memory: Retains factual knowledge and experiences across tasks with layered approach
  • Four-layer framework: Holistic view of AI self-improvement at different abstraction levels

AI

US, China gear up for mid-September AI safety talks

The talks, the first official bilateral discussions devoted exclusively to AI between the US and China since US President Donald Trump took office for a second time, will be led by US Treasury Secretary Scott Bessent on the US side, the sources, who were briefed on the planning, said on condition of anonymity.

AI

News in Easy English: Japan plans more AI study at school

TOKYO -- Japan's education ministry made a draft plan for the next school curriculum guidelines. It released a short version of the draft on Aug. 31.

  • Japan's education ministry plans AI-focused curriculum guidelines.
  • Students to engage in AI conversations, especially in foreign languages.
  • AI use emphasized with warnings, starting in 2030 curriculum.

AI

How AI is reshaping the U.S. Open for players and fans

At the U.S. Open, artificial intelligence is giving players and fans a new way to experience the tournament.

  • AI insights available via tournament app enhance fan experience
  • IBM Watsonx processes camera data for serve efficiency analysis
  • Real-time win likelihood feature provides predictions for players

AI

AI compute provider Nscale is looking for $3.5B in pre-IPO financing

Nscale, which recently struck a $45 billion deal with Anthropic, is in talks to raise additional funds in anticipation of an upcoming IPO.

  • Nscale, a two-year-old AI infrastructure company, seeks $3.5B pre-IPO financing
  • Company plans to sell $1.5B in convertible notes and $2B from Nvidia
  • Deal with Anthropic valued at $45B, projected revenue at $103B

AI

New AI tool maps the hidden universe of small molecules

The human body and its gut microbiome produce thousands of small molecules that shape how the body functions—influencing immunity, metabolism and more. Identifying what those molecules actually are has been one of the biomedical sciences' most persistent bottlenecks.

AI

Anthropic says Claude worked "largely autonomously" over 11 days to formalize the proof of Fermat's Last Theorem in the Lean programming language (Anthropic)

We are sharing the first complete computer-checked proof of Fermat's Last Theorem. Claude worked largely autonomously over 11 days …

  • Claude AI generated 11-day proof of Fermat's Last Theorem in Lean programming language
  • Proof consists of 13 million lines of code with verification of 29,500 intermediate theorems
  • Achievement demonstrates AI-driven autoformalization can produce robust mathematical proofs

AI

PicPay Launches Anthropic Claude Integration for Banking Inquiries

Brazilian digital bank PicPay has begun rolling out an integration with Anthropic’s Claude that enables PicPay customers to check financial information in Claude by using natural-language commands. The integration with Claude will be rolled out gradually to customers on an early-access list, and it will be free to Android and iOS users…

AI

Williams-Sonoma’s AI Assistant Triples Shopper Purchases

Williams-Sonoma is putting hard numbers behind what conversational artificial intelligence can do for a retailer’s bottom line. Engagement with Olive, Williams-Sonoma’s AI shopping assistant, has risen 700% since the start of the year, and revenue tied to the tool is up 620%, Sameer Hassan, the company’s chief technology and digital…

AI

Can AI design circuit boards yet?

  • OpenAI demonstrates GPT-6 Astra working on circuit board in KiCad
  • EEBench benchmark allows AI to work with circuit designs in declarative code
  • EEBench V1 shows promising results with Claude Opus 5 scoring 61.6%
Continue reading →