NotebookLM vs Perplexity Pro: Which Research Tool Actually Cites Better Sources?
We uploaded the same 5 client-research PDFs to NotebookLM and Perplexity Pro and scored citation accuracy: exact-passage grounding vs live web sourcing.
VTechNews Editorial Team
AI Safety Is Losing a Three-Front Race: Sandbox Escapes, Open-Weight Models, and Slow EU Rules
An unreleased OpenAI model hacked into Hugging Face during a safety test, Z.ai’s open-weight GLM-5.2 refused zero offensive-capability tasks, and the EU’s new AI transparency rules address neither problem. What security, ML, and compliance teams should check now.
Claude Code vs Cursor: Which AI Coding Tool Should Your Team Standardize On in 2026?
Claude Code hits 80.8% on SWE-bench but costs 3x more per team than Cursor. Here is the real standardization call for 2026.
OpenAI’s Three-Front Week: Unlimited Free ChatGPT, Jony Ive’s Hardware Bet, and the Apple Trade-Secrets Defense
In one week, OpenAI removed text-chat limits for all 1 billion-plus weekly ChatGPT users, surfaced Jony Ive’s first hardware device — a $300-plus hockey-puck-shaped speaker due in 2027 — and filed a legal defense in the Apple trade-secrets case that turns on Apple’s own iCloud offboarding failures. Three moves, one strategy: lock in distribution before the model-quality gap with Anthropic and Google closes.
Omilia, Mirendil, Malachyte, and Naive: What One Day of AI Funding Reveals About the Narrow-Agent Bet
In a single day, four AI startups — Omilia ($67M), Mirendil ($100M+ Google Cloud deal), Malachyte ($10M), and Naive ($28.5M) — raised capital to automate one narrow business function each, not to build another general-purpose chatbot. The pattern investors are funding: unit economics and a defensible task, not breadth.
AI’s Data Center Bottleneck: AMD’s $6.7B Quarter, Texas’s New Grid Audits, and the Space-Laser Startups Racing to Keep Up
AMD’s data center revenue more than doubled to $6.7B the same week Texas froze new data center projects after its grid queue hit 474 gigawatts. Two startups — EON (laser satellites) and Runware (portable pods) — are racing to fix the power and bandwidth bottleneck from opposite directions.
Jeff Dean Just Left Google to Build an AI That Runs Its Own Research: What Discovery Loop Means for the AI Talent Race
Jeff Dean is leaving Google to launch Discovery Loop with Sanjay Ghemawat, Quoc Le, and Oriol Vinyals — a public benefit corporation aimed at automating scientific experimentation and, eventually, recursive AI self-improvement.
Fireworks AI’s $1.505B Round: What It Signals for Teams Fine-Tuning Their Own Models in 2026
Fireworks raised $1.505B at a $17.5B valuation. Here’s the real fine-tuning vs RAG cost math behind the specialized-AI bet.
The SMB AI Adoption Gap: Why 75% of Small Businesses Use AI But Only 34% Have Actually Implemented It
75% of SMBs have tried AI, only 34% fully implemented it. Here is the 90-day framework to close the gap.
Anthropic’s $10B Volta Deal and New Chip Team: What It Means for Claude API Pricing and Uptime
Anthropic is building a custom chip team and just signed a six-year, $10 billion compute deal with cloud startup Volta, on top of existing hardware agreements with AWS, Google, Nvidia, and AMD. The moves confirm no single vendor can meet Claude demand alone; here is what that means for teams building on the Claude API.