AI’s moving from tabs to take-action: Opera ships an agentic browser, OpenAI tightens ChatGPT safeguards, DeepSeek slashes long-context costs, and Anthropic targets production-grade coding. Policy bookends from Sacramento to Washington meet mega-infrastructure—OpenAI’s Korea chip/data-center push—while capital pours into benefits (Judi Health), finance ops (AppZen), infra (Nscale), robotics (SwarmFarm, Kodiak), and devices (Robo.ai). Dive in for the week’s fastest movers—and what they signal next.
Industry & Enterprise
Opera Launches Neon, an AI-Centric Browser for Power Users
Opera launched Neon, a $19.99/month AI-first browser that turns browsing into agentic workflows with features like Neon Do (task automation and recall), Cards (modular prompts), and Tasks (workspaces combining AI chats with tab groups). Aimed at heavy AI users, Neon enters a crowded field alongside Perplexity’s Comet and The Browser Company’s Dia, promising deeper automation and intelligent assistance baked into the browser. (AI Insider)
OpenAI Expands ChatGPT Safety Measures With New Routing System and Parental Controls
OpenAI rolled out new ChatGPT safeguards, including per-message routing that detects emotionally sensitive conversations and directs them to GPT-5 trained on “safe completions,” following scrutiny and a wrongful-death lawsuit. New teen parental controls add quiet hours, feature blocks (voice, images), stronger protections, and self-harm detection that can alert parents and, in urgent cases, escalate to emergency services during a 120-day test period. (AI Insider)
DeepSeek Unveils V3.2-exp Model with Sparse Attention for Low-Cost Long-Context AI
DeepSeek released V3.2-exp, an experimental open-weight model introducing “Sparse Attention” with a lightning indexer and fine-grained token selection to prioritize relevant excerpts in long contexts and cut inference overhead. Early tests suggest API costs could drop by about 50% for extended-context use, positioning the approach, built on its earlier R1 work, as a practical efficiency advance that others can independently validate and potentially adopt. (AI Insider)
OpenAI Strikes Landmark Chip and Data Center Deals in South Korea to Power Stargate Project
OpenAI signed letters of intent with Samsung Electronics and SK Hynix to supply high-bandwidth memory and build AI data centers in South Korea as part of its $500B Stargate infrastructure push with Oracle and SoftBank. The plan targets up to 900,000 DRAM wafers per month—more than doubling global capacity—and includes enterprise adoption of ChatGPT Enterprise and OpenAI APIs by both conglomerates, underscoring OpenAI’s bid to lock in compute at scale. (AI Insider)
Anthropic Launches Claude Sonnet 4.5, Advancing AI for Production-Ready Software Development
Anthropic launched Claude Sonnet 4.5, a frontier coding model available in the Claude API and chatbot at the same price as its predecessor, positioned to build production-ready applications. It posted leading scores on SWE-Bench Verified and showed long-horizon autonomy in enterprise trials, while Anthropic broadened its ecosystem with a Claude Agent SDK and an “Imagine with Claude” research preview amid intensifying competition with OpenAI. (AI Insider)
Research & Innovation
How Will AI Agents Power The Future Factory? Study Maps Path to Smarter Manufacturing
AI-driven multi-agent systems integrated with MES enable adaptive, decentralized decision-making for Industry 4.0/5.0, boosting scheduling, predictive maintenance, and real-time resource allocation via machine learning, reinforcement learning, and digital twins. Adoption is constrained by scalability, interoperability, and cybersecurity gaps—and by readiness-level hurdles—necessitating industry–academia collaboration and workforce upskilling to deploy at scale. (AI Insider)
Policy & Governance
California Enacts Nation’s First AI Safety Transparency Law
California Gov. Gavin Newsom signed SB 53, the first state law requiring major AI developers (OpenAI, Anthropic, Meta, Google DeepMind) to disclose safety protocols, protect whistleblowers, and report critical AI incidents to the state Office of Emergency Services, including risks like model-enabled cyberattacks and deceptive autonomous behavior. Anthropic backed the bill while Meta and OpenAI opposed it, as California positions itself as a national model for AI oversight amid similar proposals in New York. (AI Insider)
White House Places AI as a Critical Technology in its Top R&D Priorities
The White House’s FY 2027 R&D memo puts AI at the center of federal priorities across defense, health, energy, and space, tying progress to semiconductors, secure communications, advanced computing, and manufacturing while directing agencies to build infrastructure, datasets, and evaluation methods for reliability and resilience. Crosscutting actions include “Gold Standard Science,” expanded STEM workforce development, and stronger protections against foreign exploitation alongside broader collaboration among government, academia, and industry.(AI Insider)
Startups & Capital
Capital Rx Announces Funding Round of $400M to Accelerate AI-Powered Health Benefits Platform; Rebrands as “Judi Health” to Reflect Expansion Beyond Pharmacy
Judi Health (formerly Capital Rx) raised $400M—including a $252M Series F led by Wellington Management and General Catalyst—and rebranded to reflect its evolution into a full-service health benefits technology company. Its AI-powered Enterprise Health Platform unifies pharmacy, medical, vision, and dental benefits, already serving 4M PBM members and 54M health plan lives to tackle rising costs and administrative complexity. (AI Insider)
AppZen Closes $180M Growth Round Led by Riverwood Capital to Take the Next Step in Autonomous Finance
AppZen raised $180M in Series D funding led by Riverwood Capital to speed adoption of its agentic AI platform that automates T&E, accounts payable, and corporate card workflows—shifting up to two-thirds of manual finance work to AI “digital coworkers.” Powered by ZenLM and the Mastermind platform with support for 40+ languages and complex multi-entity operations, AppZen serves 500+ enterprises (65+ Fortune 500), positioning it as a leader in autonomous finance for CFOs seeking efficiency, compliance, and better decisions without adding headcount. (AI Insider)
Australia’s SwarmFarm Robotics Raises $30 Million to for U S Expansion of its Farm Robotics
SwarmFarm Robotics raised $30 million led by Edaphon to accelerate U.S. expansion of its lightweight autonomous farm robots and partner ecosystem, with participation from CEFC ($7 million via PATF), QIC, and existing backers, according to Business News Australia. SwarmBots have demonstrated up to 95% herbicide reduction and ~35% fuel-related emissions cuts, and the platform has logged 220,000+ operating hours across 2M+ hectares, with proceeds earmarked for North American growth, hiring, and ecosystem development. (AI Insider)
Nscale Announces $433M Pre-Series C SAFE, Building on Historic $1.1B Series B Momentum
Nscale raised $433M in a Pre-Series C SAFE just days after its $1.1B Series B, with backing from Blue Owl Managed Funds, Dell, Nvidia, and Nokia. The UK-based firm runs 50+ data centers and a large GPU pipeline to deliver vertically integrated AI infrastructure—positioning itself as a sovereign, scalable hyperscaler for enterprise AI. (AI Insider)
Kodiak AI to List on Nasdaq After SPAC Merger, Raising Over $212 Million to Scale Driverless Trucking
Kodiak Robotics closed its merger with Ares Acquisition Corp II and will begin trading on Nasdaq as Kodiak AI (KDK, KDKRW) on Sept. 25, 2025, after raising $212.5M+ in PIPE and trust proceeds at roughly a $2.5 billion valuation. The company reports ~3M autonomous miles, 3,000+ paid driverless hours, 7,300+ loads, and an Atlas Energy Solutions order for 100 trucks (8 operating up to 24/7), and plans to scale its Driver-as-a-Service model with Roush Industries while marking its debut with a Times Square display and Nasdaq bell-ringing. (AI Insider)
Burkhan Capital LLC led Consortium Commits to Investing $300 Million in Robo.ai for Global AI and Robotics Platform
Burkhan Capital led a consortium committing $300M to Robo.ai (NASDAQ: AIIO), expected to close within 30 days, to accelerate growth and build a global AI-enabled smart-device ecosystem that incorporates blockchain/Web3, including DePIN via Arkreen. The deal bolsters Robo.ai’s Middle East presence and supports expansion to North America, Europe, and South Asia, aligning with Burkhan’s “profit with purpose” strategy and prior AI bets like Blaize. (AI Insider)
Check out AI Insider every day for the latest in AI an