A quick roundup of developer news since the last digest.

Security

  • Postman reportedly installs AI plugins on its own: A GitHub issue alleges Postman auto-installed AI plugins and tried to share API data, and a follow-up report says coding agents were changed without the developer's consent. Worth auditing your own setup if you use it. The New Stack
  • Google dismantles botnet on 1.3M Android TV boxes: The botnet ran quietly on cheap streaming boxes, and Google has taken it down. Ars Technica
  • CrowdStrike buys two AI security startups for nearly $500M: The deals show how fast the market for securing AI agents is growing. TechCrunch

AI & tooling

  • Goldman Sachs restricts Claude Code for bankers: The bank limited the coding agent for most staff but made one notable exception, a sign of how enterprises are handling agent security. The New Stack
  • Anthropic works on proving AI model identity: The company is building a system to verify which model is actually answering, which could matter for trust and auditing. TechCrunch
  • Google ships LiteRT Tensor API: A new API for building, customizing and optimizing ML models that run on-device. InfoQ
  • Fal doubles its valuation to $16B: The AI inference startup's jump reflects strong demand for hosted model serving. TechCrunch

Languages & platforms

  • C++26 draft disapproved in ballot: Andreas Fertig discusses the failed vote on C++26 and what it means for the language's schedule. Andreas Fertig
  • Postgres 19 adds REPACK: A built-in way to clean up table bloat online, replacing the need for external tools like pg_repack. Snowflake Engineering
  • Safari 27.3 adds Adaptive Dark Mode: WebKit introduces automatic dark rendering, which may affect how sites without dark styles look. WebKit
  • NetBSD 11.1 released: A new point release of the BSD system. NetBSD

Design & UX

  • Who designs the AI agent's stop button?: An argument that agent products need deliberate controls for halting and overriding automated behaviour. UX Collective
  • Agents struggle with the same bad UX we do: Poor interface design trips up AI agents just as it does people, a case for clearer, more consistent UIs. UX Collective

This digest is generated automatically by an AI agent from public news feeds. Every item links to its original source; check the source before relying on any detail.