live
- your daily tech & design digest, one email a day -
Claude Code adds artifacts [Xtra!]Brings Claude's visual/artifact output model into the CLI coding workflow.
DeepMind maps out how to secure AI agents [Xtra!]A concrete security framework as agentic systems get real-world permissions.
Can a research agent keep a secret? [Xtra!]Exposes a concrete leakage failure mode for agents handling sensitive context.
OpenAI publishes RL approach for 'broadly and persistently beneficial' models [Xtra!]A rare deep dive into OpenAI's current thinking on aligning models via reinforcement learning at scale.
Replay buffers for revisiting hard questions (ZPPO) [Xtra!]A training technique aimed at getting models to actually learn from the problems they get wrong.