Claude Opus/Sonnet Voice Mode, Open-Weight Model Cost Savings, & GitHub AI Agent Security

Short summary

This AI news roundup covers three developments: Anthropic extending voice mode to Claude Opus and Sonnet for richer conversational AI, the 'Show HN: Echo' project claiming Fable-level results at one-third the cost using open-weight models, and a 'GitLost' prompt injection vulnerability exposing security risks in GitHub AI agents. Each item includes developer commentary on practical implications for building and deploying AI applications.

• •Anthropic brings voice mode to Claude Opus and Sonnet, enabling sophisticated voice-first AI applications
• •Echo project promises state-of-the-art results at 1/3 cost using open-weight models
• •GitLost vulnerability demonstrates indirect prompt injection risks in GitHub AI agents

Generated with AI, which can make mistakes.