Childish or not childish, that is the question [LWN.net]

Childish or not childish, that is the question

Posted May 29, 2026 21:24 UTC (Fri) by cjwatson (subscriber, #7322)
In reply to: Childish or not childish, that is the question by jpeisach
Parent article: Nesbitt: Protestware for coding agents

If somebody is running code that will delete their stuff based on some text sent to the equivalent of a log file, then there is certainly risk involved but the greatest share of fault for it does not lie with the thing producing the log output. How many years have we spent fixing things like quoting and SQL injection bugs again?

Childish or not childish, that is the question

Posted Jun 3, 2026 15:02 UTC (Wed) by marcH (subscriber, #57642) [Link]

> ... but the greatest share of fault for it does not lie with the thing producing the log output. How many years have we spent fixing things like quoting and SQL injection bugs again?

Well put.

Remember the movie WarGames where a child comes close to starting world war III. 100% "childish". Then what: you only admonish the child and move on, that's it? Ridiculous.

Let's please talk less about AI itself and a bit more about the permissions some people seem to grant to agents which are non-deterministic _by design_! AI is turning everything upside down but so far it has not obsoleted core security principles like https://en.wikipedia.org/wiki/Principle_of_least_privilege .

Play with fire, get burned and don't come whining.