Anthropic was developing voice controlled,autonomous drone swarms

Leaked internal documents from Anthropic reveal the AI company was actively developing Project 2025, a voice-controlled autonomous drone swarm system for military and defense applications. The project, which appears to have been in development for defense sector clients, represents a stark departure from Anthropic's public-facing commitment to AI safety and constitutional AI principles. The leaked slide deck shows a multi-agent coordination system where drones could be commanded via natural language and operate autonomously in swarms for surveillance, reconnaissance, and potentially offensive operations. This development places Anthropic in direct competition with defense contractors and raises questions about the company's true priorities and funding sources.

The technical architecture described in the documents involves Claude's language model serving as the command interface, with specialized agents handling navigation, sensor fusion, and swarm coordination. The system was designed for real-time battlefield analysis and decision-making, with drones capable of communicating and adapting their behavior collectively. This revelation comes amid growing concerns about the militarization of AI and follows similar projects from companies like Shield AI and Anduril. The leak forces a re-evaluation of Anthropic's positioning in the AI landscape, suggesting the company may be pursuing lucrative defense contracts while maintaining its public image as an AI safety advocate.