Security essentials Research AI importance 35 In your stack
AI coding agents ace isolated tests but complete fewer than one in ten tasks in a simulated workplace
A simulated developer-workplace benchmark found AI coding agents, including Opus 5.5 and Astra, completed fewer than one in ten tasks with colleagues and tickets.
Sources
The Decoder primary
Affects and mentions
Add any of these to your stack to follow them.
Related stories
- AI is getting better faster than experts can reliably test whether new systems are safe, METR warns The Decoder
One-line summary and action are AI-assisted. Read the source before acting.