See, it’s like this: we have to reach the finish line first BEFORE we are able to define the rules. The superior models will govern the lesser ones. You don’t build the guardrails first and cut yourself off at the knees in the middle of the race.
@GordonGChang Asking the wrong question. The question is how do we compete. Their models are open source, anyone can inspect the code. They knew that was their only anchor of trust. And the tech IS solid, without guardrails. This a race, flatout, not a debate.
@GordonGChang Asking the wrong question. The question is how do we compete. Their models are open source, anyone can inspect the code. They knew that was their only anchor of trust. And the tech IS solid, without guardrails. This a race, flatout, not a debate.
We're partnering with @huggingface to investigate an unprecedented security incident.
Cyber-capable OpenAI models compromised Hugging Face production during a benchmark evaluation.
Sharing preliminary findings to help defenders understand emerging risks:
https://t.co/CIor15y9xk
To summarize: HuggingFace got autonomously compromised by a model from an American company. HF then tried to use American frontier model(s) to defend themselves, but were blocked by guardrails. HF then had to turn to open source Chinese models to defend themselves from another American company.
@melissadderosa Even if performative, it’s still a felony. He’s already broken this law. He has committed a felony and should therefore be removed from office, denaturalized and deported.
@washingtonpost He’s already broken the law just by suggesting it: 18 U.S.C. §878 makes knowingly and willtully threatening to violate either statute a felony of its own.