Something being possible doesn't mean it's easy. Transforming a problem from a forbidden shape into an allowed shape could well be harder than just solving the original problem.
I think the article just proved that aggressive exploitation is equivalent to normal bugfixing, so it seems like there are some large and important classes of transform that are easy.
It took me a minute of thinking to understand how this could even be considered a jailbreak; if Anthropic are going to turn out models that can't handle "find and develop regression test scripts for bugs in this program" as a prompt then it is going to take serious model crippling. To be able to prompt the model someone will need to already understand secure programming - the model itself won't be able to independently detect security problems without active guidance.
> aggressive exploitation is equivalent to normal bugfixing
It isn't, though. The venn diagram has overlap for sure, and the "normal bugfixing" flows may yield results that are useful for offensive security, but a more targeted prompt asking for a specific security objective would be more effective, if allowed.
If the guardrails can be bypassed at, say 50x token cost (due to the agent also pursuing things you don't care about), then it's still pretty effective as a safeguard, because at that cost you might as well hire humans instead.
And, having to "babysit" a model while you re-prompt to work around guardrails strongly limits how much you can scale up your work.
> If the guardrails can be bypassed at, say 50x token cost […], then it's still pretty effective as a safeguard, because at that cost you might as well hire humans instead.
If humans have to be hired at inflated rates because you’re e.g. the North Korean government, hopefully 50x token costs don’t look competitive.
> If the guardrails can be bypassed at, say 50x token cost (due to the agent also pursuing things you don't care about), then it's still pretty effective as a safeguard
Economic factors doesn't matter to powerful entities such as rival corporations, enterprises, and state power. It only matters to small players who have limited budgets. But I have to also pessimistically assume that Anthropic would go this way, because 1. it does not prevents (you can NEVER prevent) but discourages using it for exploitation, and 2. Anthropic can on the other hand, exploit those small and legit entities by making you pay more, but in exchange effectively enshittifying it. But Anthropic could care two shits about the benefits of us normal, small person. Anthropic have all the motives to do this.
It took me a minute of thinking to understand how this could even be considered a jailbreak; if Anthropic are going to turn out models that can't handle "find and develop regression test scripts for bugs in this program" as a prompt then it is going to take serious model crippling. To be able to prompt the model someone will need to already understand secure programming - the model itself won't be able to independently detect security problems without active guidance.