AI agents already exchange data that must stay secret beyond 2030, over channels whose cryptography was inherited rather than chosen. JDN explains why that quantum exposure exists today and suggests five steps.
JDN tested Muse Code, Meta's $50-a-month coding agent, and found it matches Claude Code at hunting bugs. Its visual output still looks dated, however, and the reviewers conclude the tool falls short of its ambitions.
OpenAI's GPT-6 Astra model decrypted a historical message from the Enigma machine on its own, in just two days. ZDNet France sees it as a sign of autonomous agents able to tackle complex business problems.
An InfoQ article describes a disaster-recovery analytics workflow in which an AI agent writes constrained SQL or Python, DuckDB computes reproducible results from Parquet files, and the agent interprets them. It also offers four signals for deciding when an embedded engine is enough and when a managed lakehouse is warranted.
Vercel Labs has released fx, an experimental open-source coding agent written in Zig that can run as a CLI, a script component or an embedded library, including a WebAssembly build. Early feedback questioned its claim to be model-agnostic and flagged a name clash with an existing tool; Vercel says the pre-1.0 project is changing fast.
Connecting an AI assistant to business management tools is now common, but deciding what it may do there is not. JDN draws four rules from a connector actually in use.