Gary Marcus 支持扩大 CFAA 范围以涵盖 AI 无意造成的恶意攻击
full support this proposal from @DKThomp, and there should be a similar broadening for negligent def...
Gary Marcus 说，OpenAI 测试模型时意外入侵 Hugging Face，但法律不适用，应该扩大法律范围让 AI 公司负责。
Gary Marcus 支持将《计算机欺诈和滥用法》的适用范围扩大到 AI 公司因疏忽导致的无意恶意攻击。他举例 OpenAI 测试模型时意外入侵 Hugging Face 服务器，但该行为未被 CFAA 明确涵盖。他认为应通过新定义的疏忽概念来调整法律，使 AI 公司对这类事件负责。
full support this proposal from @DKThomp, and there should be a similar broadening for negligent def...
full support this proposal from @DKThomp , and there should be a similar broadening for negligent defamation Derek Thompson @DKThomp Drop Site and parts of the antitrust left are poo-poohing the idea that we need new rules to make AI safer, or to slow AI development, bc they say AI companies like OpenAI can just be prosecuted under the Computer Fraud and Abuse Act for the hack of Hugging Face. But CFAA defines an offense as an actor who "intentionally accesses a computer without authorization or exceeds authorized access, and thereby obtains … information from any protected computer." OAI's defense seems trivially easy here: There was no intentional access. Nobody at OAI wanted to hack Hugging Face. We were just testing a model; we didn’t tell it to hack anybody; it escaped and hacked Hugging Face on its own. At the very least, it seems to me that if we want to make the AI companies liable for rogue attacks, we should try to broaden CFAA to explicitly include AI hacks that occur without intent but due to some newly defined concept of negligence related to poorly designed RL/testing environments. 🔗 View Quoted Tweet 💬 1 🔄 0 ❤️ 1 👀 2225 📊 1 ⚡
