Anthropic urges a global pause over self-improvement risk
Anthropic warned that AI systems will soon improve themselves without human intervention, a potential danger marker. Critics called it a marketing tactic or a way to slow competitors.
Anthropic says 80% of its new production code is authored by Claude
Framed as a signal that matching it requires shifting from a "developer assistant" to an "automated factory" architecture.
TLDR →
Productivity · Agents
Anthropic
Apple approves the first AI agent for Messages Business Chat
The Poke assistant arrives on iMessage via Apple Messages for Business, able to send emails, set reminders and generate images, integrating with Oura, Outlook, Gmail, GitHub and Navan, with light actions free and heavier requests paid.
TLDR →
Consumer AI · Agents
Apple, Poke
Amazon's Proteus warehouse robot goes fully autonomous
An AI upgrade lets workers direct Proteus in plain language while it works out priority, route and timing. Piloting in Amazon labs, with European deployment targeted for the first half of 2027.
Open Code Review released by Alibaba
An AI-powered code-review CLI that reads git diffs, sends files to a configurable LLM via a tool-using agent and produces line-level review comments.
Bots overtake humans in web traffic
AI agents performing tasks for people have tipped the bot-versus-human balance of online traffic.