Gary Marcus 评论 Blumenthal 致 OpenAI 质询信与 Protect Democracy 透明度诉讼

Pushing for transparency

1. Senator Blumenthal just sent a great letter to OpenAI, reminiscent of the famous Watergate-associated phrase, “What did the President know, and when did he know it?”. The letter demands detailed answers about OpenAI’s agents, their role in various hacks, and when OpenAI found out. The company’s lack of transparency is deeply worrisome, and I am glad the letter is pushing them for the facts.

   

Click here for the Senator’s full list of questions.

1. A few days ago, Protect Democracy filed an important lawsuit to push the Trump administration to be more transparent about the criteria by which AI systems are being evaluated.

   

Excerpt from Protect Democracy’s complaint As I have argued here multiple times, the current lack of transparency in U.S. AI regulation actually hurts American businesses by adding unpredictability (potentially leading people to use software from other vendors) and gives the appearance of possible favoritism and/or corruption. Opacity also becomes a potential tool for authoritarianism.

Meanwhile, the current opacity keeps the scientific community from weighing in on the adequacy (or inadequacy) of the current criteria. I am worried, for example, that the criteria may pay insufficient attention to the importance of monitorability, one of the few tools we have to keep an eye on what AI agents are doing. In particular, GPT-6 Astra has less monitorability, perhaps increasing risks — yet the government apparently waved the model through.

I wish Protect Democracy the best in their suit. We will all be better off if they win.

It’s heartening to see both efforts, from Senator Blumenthal and from Protect Democracy. If we are going to address the risks that Anthropic and OpenAI’s current and former staff have been raising, the least we can do is find out what the hell is actually going on.

Your support keeps me going; please consider becoming a subscriber.