The Tech DocketDaily signal on tech & AI — what the world is searching, and why

Illustration of a code editor with a context gauge being dialed down from 372K to 272K tokens

AI Tools & Apps · July 19, 2026 Codex Context Window Cut to 272K: What OpenAI Rolled Back

OpenAI cut Codex's default context from 372K back to 272K tokens, matching its own pricing cliff. What changed, why, and how Claude and Gemini compare.

Latest

Flat illustration of a smartphone chat bubble connected to a cloud server behind a padlock, representing what happens to AI chatbot conversation data

AI Tools & Apps · July 15, 2026

AI Chatbot Data Privacy: What Happens to Your Chats

ChatGPT, Gemini, Claude and Meta AI set different defaults for training on free chats, retention and opt-outs — compared using each vendor's own policy pages.

Stylized graphic card reading Bonsai 27B, a 27-billion-parameter AI model compressed to run on a smartphone

AI Models · July 15, 2026

Bonsai 27B: The 27B AI Model That Runs on a Phone, Explained

PrismML's Bonsai 27B squeezes a 27B-parameter AI model into 3.9 GB to run on flagship phones. What's verified, what's hype, and which phones in India can run it.

A data center aisle of GPU servers with a rising electricity meter and dollar-per-token price tags overlaid on the racks

Chips & Hardware · July 15, 2026

The Real Cost of Running AI: Tokens, GPUs, Power

OpenAI, Google and Anthropic's live per-token prices, GPU purchase and rental costs, and IEA electricity data, lined up to show what AI really costs to run.

Split-screen comparison of a real photograph and an AI-generated image with a Content Credentials verification icon overlaid

How-To & Explainers · July 15, 2026

How to Spot AI-Generated Images and Videos in 2026

A practical, source-checked guide to Content Credentials, SynthID, platform AI labels, detector accuracy limits, and India's 2026 deepfake rules.

All articles →