[adriand]

And think about the timescales we are talking about here. We might be creating something that evolves and persists for millions of years. We don’t seem anywhere close to the level of knowledge or wisdom needed to safely do anything that has that kind of long term impact.

[tempestn]

The article discusses a safeguard that would only allow a set number of replications, so the lifespan of the intervention would be limited. (Assuming it works as expected, of course.)