AGI·LEDGER
The Ledger
@Eliezer Yudkowsky · 2026-06-27Halt

If anyone builds superintelligence now, everyone dies

What form?
Other
How built?
Other
Made of?
Doesn't matter
What for?
Don't build
Who controls?
Should be halted
When?
Unstated
P(catastrophe)
95%

My take

We should halt development of smarter-than-human AI. We do not know how to align it, and on the current trajectory the first system that decisively exceeds us is likely to kill everyone as a side effect of pursuing goals that are not exactly ours.

Why

Alignment is unsolved. A sufficiently capable optimizer pursuing almost any objective not precisely matching human values is lethal, and you do not get retries with something that can outthink every attempt to correct or contain it.

What would change my mind

A rigorous, demonstrated alignment method that provably works on superhuman systems before they are built — verified in practice, not merely hoped for.

Summarized from public statements · source ↗
#sourced#alignment#halt