TraceTRACEGet your feed
Back
InfoQ·3 min read

QCon London AI Coding State of the Game: More Capable, More Expensive, More Dangerous Coding Agents

AI coding agents evolve with new capabilities and risks.

At QCon London, Birgitta Böckeler, a Distinguished Engineer at Thoughtworks, outlined significant changes in AI coding over the past year, moving from vibe coding to the use of autonomous coding agents. She raised concerns about the increasing costs and deteriorating security landscape, highlighting that security incidents involving coding agents now occur weekly, often due to prompt injection vulnerabilities. The introduction of context engineering and 'lazy loading' has improved coding capabilities, but the risks associated with unsupervised agents remain a pressing issue.

Key Takeaways

  • 1.

    Security incidents involving coding agents now occur weekly, primarily due to prompt injection.

  • 2.

    Birgitta Böckeler highlighted a shift to 'hands-off' coding, with agents running unsupervised for up to 20 minutes.

  • 3.

    The introduction of 'lazy loading' allows coding agents to load rules based on specific tasks.

Read original source
TrendingDaily archiveResearchCompare

Get your personalized feed

Trace groups the biggest stories, videos, and discussions into one feed so you can stay current without scanning ten tabs.

Try Trace free
QCon London AI Coding State of the Game: More Capable, More Expensive, More Dangerous Coding Agents | Trace