Anthropic has come out against a proposed Illinois law backed by OpenAI that would shield AI labs from liability if their systems are used to cause large-scale harm, like mass casualties or more than $1 billion in property damage.