It was a week of big launches, partnerships and another exodus in the ongoing talent war. Anthropic pushed Claude Code to the web, Adobe launched AI Foundry for brand-tuned models, and IBM teamed with Groq to speed agentic AI—while Apple lost another AI lead to Meta and Meta trimmed roles amid a lab reorg. DeepMind paired with Commonwealth Fusion Systems to apply AI to fusion control, signaling AI’s reach into next-gen energy.
On the policy and research side, Waterloo unveiled a math-plus-ML framework to verify AI safety and the UK earmarked £8.9M to speed approvals for drones, robotics, and AI tools. Startups stayed hot, with OpenEvidence’s $200M raise at a $6B valuation, Romania’s .lumen landing €11M for delivery robots, and fresh capital for Periodic Labs, Arcjet, and Resistant AI.
Industry & Enterprise
Anthropic Launches Web App for Claude Code, Expanding Access to Its AI Coding Assistant
Anthropic launched a web-based Claude Code for Pro and Max subscribers, letting developers create and manage multiple AI coding agents from the browser and iOS, expanding beyond the original CLI. Adoption has grown 10x since May to $500M+ in annualized revenue, positioning Claude Code as a leading rival to Copilot, OpenAI, and Google as Anthropic deepens cross-platform integration. (AI Insider)
Adobe Launches AI Foundry to Help Enterprises Build Custom Generative AI Models
Adobe launched AI Foundry, an enterprise service that fine-tunes Firefly models on a company’s own branding and IP to generate on-brand text, images, video, and 3D content, with usage-based pricing. Adobe says enterprises have created 25+ billion assets with Firefly since 2023, positioning Foundry to deepen brand-specific workflows while augmenting—rather than replacing—creative teams. (AI Insider)
IBM and Groq Partner to Accelerate Enterprise AI Deployment
IBM and Groq are integrating GroqCloud’s ultra-fast inference into IBM’s watsonx Orchestrate to power agentic AI at enterprise scale, promising lower latency, better scalability, and improved cost efficiency for sectors like healthcare and finance. The roadmap adds support for IBM Granite models and Red Hat’s vLLM to streamline deployment and speed adoption in mission-critical workflows. (AI Insider)
Apple’s AI Search Leader Ke Yang Departs for Meta Amid Ongoing Talent Exodus
Apple lost Ke Yang, the AI lead for its web search efforts, to Meta, adding to a string of high-profile exits from its AIML unit amid a coming March Siri overhaul. The continued turnover—following departures like Ruoming Pang and others to Meta’s Superintelligence Labs—raises doubts about Apple’s ability to retain top AI talent as rivals accelerate AI search and assistant development. (AI Insider)
Commonwealth Fusion Systems Partners with Google DeepMind to Apply AI in Advancing Fusion Reactor Control
CFS partnered with Google DeepMind to apply Torax AI to plasma simulation and control for the SPARC fusion reactor, aiming to sustain stable reactions and accelerate path to net energy. Backed by Google’s investment and a 200 MW offtake from CFS’s first commercial plant, the deal underscores the AI–clean energy convergence as SPARC targets first net energy in 2026. (AI Insider)
Meta Cuts 600 Positions in AI Division as Part of Superintelligence Lab Reorganization
Meta will cut about 600 roles in its superintelligence lab to streamline decision-making and concentrate impact, with most affected employees expected to be reassigned internally. The move follows rapid AI hiring—including 50+ researchers from rivals—and aligns with Mark Zuckerberg’s “year of efficiency” push to keep Meta competitive in advanced AI. (AI Insider)
Research & Innovation
Researchers Develop System to Verify Safety and Stability of AI-Driven Control Systems
University of Waterloo researchers built a math-plus-ML framework that learns Lyapunov functions and uses logic-based checks to verify the safety and stability of AI controllers for critical systems like grids and autonomous vehicles. Published in Automatica with TRuST support, the approach automates safety proofs and will be released open source to help engineers harden real-world AI. (AI Insider)
Policy & Governance
UK Eases Rules to Speed New Tech Into Use, Backed by £8.9M for 16 Projects
The UK awarded £8.9M via the Regulatory Innovation Office to 16 projects aiming to cut red tape and speed drones, AI tools, and robotics from lab to market. Backed by RIO (chaired by Lord Willetts), pilots span London Fire Brigade risk-assessment apps, Milton Keynes street-sweeping robots, and expanded medical-drone/space initiatives, with partnerships—including an IBM AI hackathon—designed to streamline approvals and unlock multibillion-pound markets by 2030.. (AI Insider)
Upcoming Conferences & Events
GovAI Summit
Oct. 27–29, Arlington, Va., GovAI Summit convenes federal, state, and local leaders with AI innovators and infrastructure providers to align strategy, capabilities, and execution for public-sector AI. Positioned as the first major forum for implementing the 2025 American AI Action Plan, it offers hands-on programming, trusted collaboration, and policy context on infrastructure, workforce, and secure-by-design systems. (GovAI)
ROSCon 2025
Oct. 27–29,Singapore, ROSCon 2025 offers three days of workshops, technical talks, tutorials, panels, lightning talks, demos, and networking for ROS developers from beginner to expert. Modeled on PyCon/BoostCon, the fourteenth ROSCon focuses on practical tips, new tools and libraries, and community knowledge-sharing across the ROS ecosystem. (ROSCon 2025)
Startups & Capital
OpenEvidence Closes $200M in Funding at $6B Valuation to Expand AI Platform for Medical Professionals
OpenEvidence raised $200M at a $6B valuation, just three months after a $210M round, signaling intense investor appetite for specialized medical AI. The 2022-founded “ChatGPT for doctors” trains on top journals and now powers 15M+ monthly clinician queries via an ad-supported, free platform for verified professionals, with backing from GV, Sequoia, Kleiner Perkins, Blackstone, Thrive, Coatue, Bond, and Craft. (AI Insider)
SehaTech Secures $1.1M Seed Round to Accelerate AI-Powered Health Insurance Automation in Egypt
SehaTech raised $1.1M in Seed funding—bringing total backing to $2M—to expand its AI-driven platform that digitizes and automates health insurance administration across Egypt and emerging markets. Led by Ingressive Capital with Plus VC, A15, Beltone Venture Capital, and others, the round will fund team growth, enhanced AI capabilities, and regional scaling to boost access and reduce fraud in healthcare financing. (AI Insider)
Romanian Startup .lumen Gets €11M Grant to Develop Autonomous Delivery Robots
.lumen won a €11M grant for “PABLO” (2025–2028) to develop humanoid and quadruped sidewalk delivery robots with a Romanian consortium. The 56.86M RON project extends PAD AI to last-mile logistics, aiming to cut costs and emissions while delivering a full software-hardware stack, new R&D roles, and IP. (AI Insider)
OpenAI and Google Brain Veterans Launch Periodic Labs with $300M Seed Round to Pioneer AI-Driven Scientific Discovery
Periodic Labs emerged from stealth with a $300 million seed led by Andreessen Horowitz, founded by OpenAI’s Liam Fedus and Google Brain’s Ekin Dogus Cubuk, with backing from Nvidia’s NVentures, Accel, General Catalyst, DST, Felicis, and angels including Jeff Bezos, Eric Schmidt, Elad Gil, and Jeff Dean. The San Francisco startup aims to fuse large language models, robotics, and physics simulations into an autonomous discovery stack—starting with next-gen superconductors—staffed by two dozen-plus elite researchers across AI, materials science, and quantum physics. (AI Insider)
Arcjet Raises $8.3M to Tackle Surge in AI-Powered Cyberattacks and Malicious Bots with In-Code Defense
Arcjet raised $8.3M Series A led by Plural and Ott Kaukver (with a16z, Seedcamp, others), bringing total funding to $12M to scale its in-code, AI-powered threat detection platform. Already protecting 500+ production apps, Arcjet’s real-time defenses target adaptive attacks without slowing deployment, positioning it as a next-gen security layer for AI and e-commerce. (AI Insider)
Resistant AI Raises $25M in Series B Funding to Empower AI Agents to Fight Fraud and Fincrime
Resistant AI raised $25M Series B led by DTCP, with Experian, GV, and Notion Capital joining, to scale its fraud detection, transaction monitoring, and threat intelligence globally. Positioned as a profitable EU leader with 10x ARR and 150M documents verified, the company targets APP fraud, synthetic IDs, money laundering, and gen-AI forgeries as financial crime grows more sophisticated. (AI Insider)
Check out AI Insider every day for the latest developments in AI and robotics.
