We have serious proof that these models are capable of sophisticated attacks.
So what’s the trick in this honestly? OpenAI shouldn’t care about alignment and just release it? I’m genuinely asking.
Why would a company announce that their product is so good they aren’t releasing it? It’s marketing their other models at best, and at worst they’re trying to trick politicians into regulating the f-ck out of AI and make them the primary trustees à la cell carrier pseudo-monopolies.