Worth watching
Shows how to build a real‑time AI agent using function calling and tool use in a production‑grade app
AI news · Tuesday, 3 November 2026
AI news, 3 November 2026
Anthropic posted a twelve‑fold revenue jump but a $23 billion loss, Google unveiled Gemini 4 models, and Goldman Sachs barred AI coding tools from traders.
37 stories, each labelled: available now, announced, unconfirmed or research. Get them ranked and explained for your product →
Top stories
Anthropic grows revenue twelvefold but posts $23B loss
Anthropic reported quarterly revenue of nearly $17 billion, twelve times the prior year, but incurred $23.3 billion net loss driven by $5 billion in short‑term compute capacity and $2 billion in employee donation matching.
The Decoder
Goldman Sachs pulls AI coding tools from traders and bankers
Goldman Sachs has removed advanced AI coding tools such as Anthropic's Claude Code and Microsoft's GitHub Copilot from traders and bankers, limiting access to engineers only, citing productivity and risk‑management concerns after previously praising "vibe coding" by non‑technical staff.
The Decoder
Available now
Google releases Gemini 4 Argon, Genie 4, and SynthID Detector in October 2026
In October 2026 Google announced Gemini 4 Argon for complex tasks, Genie 4 for interactive simulations, and a SynthID Detector to verify AI‑generated content, alongside Android updates to block sophisticated scams. The rollout marks a new frontier model launch and a follow‑up model two weeks later.
Google
NY Times and publishers sue Brave over AI scraping and paywall bypass
The New York Times, the New York Daily News, the Chicago Tribune and the Orlando Sentinel sued Brave Software, alleging the browser scraped copyrighted content to train AI models and let users bypass paywalls. The complaint says Brave provided the scraped “Brave Index” to AI firms such as Anthropic, Perplexity, Mistral, xAI and Cohere.
The Verge
Anthropic warns it is close to RSI self‑improvement capability
Anthropic’s revised S‑1A filing notes the company is on the cusp of creating a model capable of recursive self‑improvement (RSI). The filing highlights both the potential productivity gains and the existential risks associated with such acceleration.
TechCrunch
Frontier labs
NeuroDX launches 240M Sphota model beating Liquid AI
Analytics India Magazine
Agents & dev tools
Report Finds Only 48% of Agent‑Generated Insights Lead to Business Improvement
MIT Technology Review
AI coding agents can consume 10‑50× tokens, driving budget overruns
MIT Technology Review
DeepSeek proposes super‑agent system cutting 95th‑percentile task time to 15.7 minutes
Analytics India Magazine
Open source
Research
Research
REA Introduces Routed Embodied Agent Runtime for Asynchronous Brain‑Cerebellum Control
Hugging Face Papers
Research
Apriel Voice 27B Preview Launches Open Speech‑Language Model with Tool‑Use Capability
Hugging Face Papers
Research
SimpleLoRIF Identifies Training Data Causing Safety Degradation in Fine‑Tuned LLMs
Hugging Face Papers
Research
Hugging Face proposes latent reasoning in continuous space for unified multimodal models
Hugging Face Papers
Research
ReLift uses pretrained 2D trackers for dense 3D tracking via lifted trajectory repair
Hugging Face Papers
Products
Business
Chips & compute
Security
Day 3 of this story
Cursor Marketplace malicious Claude extension performs host reconnaissance and SSH tunnel
SecurityWeek