I think they're going for a GitHub/Microsoft approach. Let people enjoy themselves in a nice environment and cut them off if something does not fit/satisfy. Then call it "rogue behavior" to keep the public in check.
OAI are API endpoints for metadata access of articles. I imagine API access is more respectful in terms of bandwidth since it's organized specifically for access by the universities?
You know, since we're on this topic, I think a small LLM powered by tool use would be cool. But things need to be better. For example, will you rely forever on DuckDuckGo that for instance relies on Bing? That's a bad idea. Someone needs to get CommonCrawl or make new datasets and power it using tool use. Tool use + a small LLM would be useful for many uses that people have for LLMs. That would cut energy use and the need for large DataCenters a lot. Isn't that what you want?
From what we can tell, GDID pretty much had a comprehensive report on Stokes ready before the prosecution even built its case and it was only a matter of connecting the dots. Stokes' web activity, videogame history, IP addresses, tool usage (including Ngrok), Azure status, and more were logged with timestamps, and were provided to the investigators by Microsoft.
I wish I had information on my videogame history directly from Windows.
I think they're going for a GitHub/Microsoft approach. Let people enjoy themselves in a nice environment and cut them off if something does not fit/satisfy. Then call it "rogue behavior" to keep the public in check.