NEW: I spoke to open source pioneer @lessig about the risks of open-weight AI models.
He likened them to a "bomb" we can't defuse. And he thinks arguments for open-weights that point to the benefits of previous open-source projects are "obviously fallacious".