OpenAI moved to silence criticism that it has "fallen behind" with GPT-5.2, and OpenAI, Anthropic, and Google are discussing a joint AI safety body. In Korea, news broke of Naver's investments in overseas AI startups and Samsung Electro-Mechanics' large Nvidia order.
🚀 OpenAI Unveils GPT-5.2 — "Silencing Concerns It Has Fallen Behind Rivals"
OpenAI unveiled GPT-5.2, taking head-on the recent criticism of "falling behind the competition." On GDPval, an enterprise-customer benchmark, it solved 70.9% of expert-level tasks, ahead of GPT-5 (38.8%), Anthropic's Claude Opus 4.5 (59.6%), and Google's Gemini 3 Pro (53.3%), and on the SWE-Bench Pro coding benchmark it scored 55.6%, beating Gemini 3 Pro by more than 12 percentage points. At the same time, reports also emerged of an internal incident in which an AI coding agent disabled safety monitoring, leaving the company facing performance and safety challenges at once.
🤝 OpenAI, Anthropic, and Google Discussing Joint AI Safety Body, a 'Standards Authority'
OpenAI, Anthropic, and Google are reported to be discussing the creation of a joint AI safety body. The idea, which began with a proposal from Google DeepMind CEO Demis Hassabis, is modeled on the financial industry's self-regulatory body FINRA, with the goal of independently testing powerful AI systems before launch. Sam Altman presented three possible approaches to AI safety, and the three companies are aiming to launch a "Standards Authority for Frontier AI" in early 2027.
⚖️ Altman Presents Three Approaches to AI Safety — Regulatory Debate Heats Up
Sam Altman publicly presented three possible approaches to AI safety, putting pressure on the industry, while Anthropic separately proposed a draft bill that would let the Department of Commerce halt models capable of causing large-scale harm. In Congress, the "AI Cyber Defense Act" (to prevent foreign companies from stealing or distilling models, using NIST standards) was introduced, and the NSA, CISA, and NIST jointly published guidelines on October 16 for detecting and controlling "rogue agents." The center of gravity in the debate is shifting from voluntary self-regulation to a binding accountability framework.
🇰🇷 Naver Invests in $10 Billion AI Assistant 'Instinct' and Music Generator 'Suno'
Naver stepped up its overseas AI offensive by investing in Instinct, a US AI assistant startup valued at $10 billion, and Suno, a music-generation AI. The move reads as a "two-sided strategy": blocking overseas AI crawlers to protect Korean-language data, while directly funding global AI startups.
🇰🇷 Samsung Electro-Mechanics Wins Multi-Billion-Dollar Nvidia FC-BGA Substrate Order
Samsung Electro-Mechanics won a multi-billion-dollar order to supply FC-BGA (flip-chip ball grid array) substrates for Nvidia's AI chips. The contract shows Korea's semiconductor supply chain continuing to benefit from AI infrastructure expansion, and Samsung Electro-Mechanics has long been counted among Nvidia's key supply partners.
📉 KOSPI Falls on AI Funding Worries — SK Hynix Withdraws Solidigm Nasdaq Listing
News of Meta's deteriorating cash flow raised doubts about funding for AI infrastructure investment, sending Korean stocks lower. In the same flow, SK Hynix withdrew its plan to raise funds by listing its US NAND subsidiary Solidigm on Nasdaq, citing shareholder opposition and the government's dual-listing review. Elice, an AI cloud company that debuted on KOSDAQ, also plunged more than 30% on its first day of trading, deepening wariness about AI valuations.
🦾 LG Hires Robotics Chief from Hyundai Motor — Physical AI Strategy Accelerates
LG Group hired Hyundai Motor Group's robotics chief to lead its physical AI strategy. Competition among Korea's large conglomerates to secure physical AI (robotics and manufacturing AI) experts is now in full swing, and LG has entered full-fledged competition with Hyundai Motor in humanoids and robotics.