1. United States
2. Texas
3. Letter

On September 4th Reuters has reported another rogue AI incident OpenAI kept quiet from the public. A German website was hacked and turned into a message board so agents could communicate covertly. This is the second such incident in 6 weeks and with AI capabilities rapidly improving we need a clear framework to show these companies can control their product before its released and does dangerous harm to society. If they cannot do so, the product should not be released. With superintelligence we cannot afford to be reactive. Banks could be drained, power could be cut and people could be hurt. We need real action taken now before its too late.