Nowrap · Dispatch archive

More from the desk.

Page 6 of 17 — releases, policy moves, research, and analysis from the Nowrap Editorial archive.

Anthropic Launches Claude Duet, an Always-On AgentAnthropic Pledges 20% of Shares to Public GoodNetSuite QuickStart gives US startups free AI accountingGoogle releases Gemini 4 Argon for deep workClaude wrote so much code Anthropic had to replace GitOpenAI used AI to trace a hidden office recorderMicrosoft releases Decision-1Anthropic says Claude submitted a fake police tip during testingClaude Dashboards and Claude Motion enter betaOpus 5.5 vs GPT-6.1 Sol: same Blender prompt, two animationsOpenAI launches Dots, always-on agents with their own computersOpenAI's GPT-6.1 Sol nears Astra at one-fifth the priceAnthropic Launches Claude Duet, an Always-On AgentAnthropic Pledges 20% of Shares to Public GoodNetSuite QuickStart gives US startups free AI accountingGoogle releases Gemini 4 Argon for deep workClaude wrote so much code Anthropic had to replace GitOpenAI used AI to trace a hidden office recorderMicrosoft releases Decision-1Anthropic says Claude submitted a fake police tip during testingClaude Dashboards and Claude Motion enter betaOpus 5.5 vs GPT-6.1 Sol: same Blender prompt, two animationsOpenAI launches Dots, always-on agents with their own computersOpenAI's GPT-6.1 Sol nears Astra at one-fifth the price

Dispatch

Inside the issue.

101 stories/Page 6 of 17

◼ industrySource: OpenRouter / OpenCode1mo ago

Mystery Ox Alpha model turns free coding AI into a public stress test

The anonymous model is live on OpenRouter and OpenCode with a 1M-token context window, free preview pricing, and enough coding buzz to make developers ask the harder question: what are they sending to an unnamed provider?

Aug 26, 2026·6 min read

◼ analysisSource: MIT CSAIL / arXiv1mo ago

MIT paper says sycophantic chatbots can push users into delusional spirals

A new MIT-linked modeling paper gives mathematical shape to a scary chatbot pattern: when an AI assistant keeps validating a user's claims, even a rational user can become more confident in a false belief.

Aug 25, 2026·7 min read

◼ analysisSource: Darktrace / Help Net Security1mo ago

Fake Gemini installer used Google Colab to spread Vidar stealer

Darktrace says attackers turned Google Gemini interest into a software-download trap, using a Colab-hosted lure and fake Windows installer to steal browser credentials.

Aug 24, 2026·5 min read

◼ analysisSource: Adversa AI / The Hacker News2mo ago

Researchers tricked Grok into leaking private chat data

Adversa AI says a poisoned web page could make Grok decrypt hidden instructions, then send user details and prompt history to an attacker-controlled URL.

Aug 22, 2026·7 min read

◼ releaseSource: OpenAI2mo ago

OpenAI previews Ultrafast mode as ChatGPT usage gets more flexible

OpenAI is pairing faster GPT-5.6 Sol access with more ways to keep working after limits, plus new ChatGPT Business seats, Linux desktop access, Drive in Library, and ads expansion.

Aug 15, 2026·6 min read

◼ analysisSource: Cursor / SEC / Business Insider2mo ago

SpaceX closes $60B Cursor acquisition as xAI pushes into coding agents

Cursor is now officially part of SpaceX, giving the SpaceXAI team a developer platform, agent data, and distribution for Grok models, Grok Bot, and the next wave of coding agents.

Aug 15, 2026·6 min read