AI labs as Jurassic Park: Two posts disagree on the safeguards

Using Jurassic Park as an analogy, one post argues that costly safety work at AI labs could make safeguards cheaper and more standardized for later entrants. It says labs would have to publish safety cases explaining which alignment and containment standards work at a given capability level. The author argues this could buy time and prevent avoidable mistakes, while acknowledging the possibility of catastrophic failure. Another post disputes the picture of careful oversight: “The ‘i’s are not getting dotted.” It says some people who could help are not involved.