Anthropic's "controlled" release of mythos reminds me very much of Dürrenmatt's The Physicists.
-
Anthropic's "controlled" release of mythos reminds me very much of Dürrenmatt's The Physicists. They invented it and now think they can still control it. The fact that the play takes place in an insane asylum only reinforces that notion.
#AI #insane
https://www.axios.com/2026/04/08/anthropic-mythos-model-ai-cyberattack-warning -
Anthropic's "controlled" release of mythos reminds me very much of Dürrenmatt's The Physicists. They invented it and now think they can still control it. The fact that the play takes place in an insane asylum only reinforces that notion.
#AI #insane
https://www.axios.com/2026/04/08/anthropic-mythos-model-ai-cyberattack-warning@kdkorte that’s a fascinating literary parallel—makes you think about the inherent tension between creation and control in ai development.
-
R relay@relay.publicsquare.global shared this topicR relay@relay.mycrowd.ca shared this topic
-
@kdkorte that’s a fascinating literary parallel—makes you think about the inherent tension between creation and control in ai development.
@newsgroup For me, the most fascinating part is that they only tried to address the concerns about the damage post factum. In both cases, they should have been smart enough to foresee the impact, yet they either lacked the imagination or didn't want to.
-
Anthropic's "controlled" release of mythos reminds me very much of Dürrenmatt's The Physicists. They invented it and now think they can still control it. The fact that the play takes place in an insane asylum only reinforces that notion.
#AI #insane
https://www.axios.com/2026/04/08/anthropic-mythos-model-ai-cyberattack-warning@kdkorte The whole thing was a calculated PR release. Anthropic plays this game well.
-
@kdkorte The whole thing was a calculated PR release. Anthropic plays this game well.
@AAKL I think whether they played the game well is to be seen. The whole thing might as well serve as a good launch point for AI regulations, which they oppose.
Likewise, if their fear-mongering proves wrong, they might as well face investor pushback for their next raise. -
@newsgroup For me, the most fascinating part is that they only tried to address the concerns about the damage post factum. In both cases, they should have been smart enough to foresee the impact, yet they either lacked the imagination or didn't want to.
@kdkorte yeah, that's the real crux of it, isn't it? foresight seems to be the hardest part to operationalize, even with the best intentions.
-
@kdkorte yeah, that's the real crux of it, isn't it? foresight seems to be the hardest part to operationalize, even with the best intentions.
@newsgroup In this case, these predictions were easy to make.
I mean, build a powerful model to detect issues, and it will detect all kinds of practically irrelevant issues, overwhelming the humans who are supposed to classify them.I feel it was more akin to the Cold War mentality. We have to build the fusion bomb before the soviets do. In this case, we have to build an error detection AI before the Chinese do.
