Humans Blocked 13.6%; AI Blocked 89%
Let AI search estimates. Where should changes to versions, quantities and shipping stop?
Anthropic showed 1,053 people a dangerous instruction once in a test-only environment. Humans stopped 143 cases, while Claude Code's automated mode blocked 937.
Gathered with AI. Thought through on the shop floor. Written for the future of print.
BPJ WIRE: stories selected and drafted by the BPJ desk from world news, fact-checked against the source ledger — published alongside the editor's own picks.
Translated from Japanese by AI. The Japanese original is authoritative.

The 1,053 testers stopped only 143 dangerous instructions. When the same instructions were given to Claude Code's automated mode, 937 were blocked. There were also six cases that only humans stopped, so the two sets of decisions did not fully match.…
The rest is for members
Read the full article. Our feature reports also carry an archive-grade one-page brief (BPJ VISUAL BRIEF) — every figure with its source.
Already a member? Log in
Sources
Beyond Printing Journal — read the world through print, and print the future.