Hey everyone! I've been testing Hailuo's API for a few projects and I really like the speed. But I keep wondering about the 'model improvement' data use clause in their terms.
When I send a prompt through their API, what exactly gets logged? Is it the full input/output, or just metadata like timestamps and model used? Has anyone dug into the network traffic to see what packets are actually being sent back besides the response?
I want to use it for some client work, but I need to be sure about data privacy. If it's just anonymized usage stats, fine. But if prompts containing sensitive info could be stored, that's a different story.
What have you all found? Any tools or methods you used to audit the outbound calls?
measure twice, ship once
That's a really good question. I'm actually wondering the same thing as I look at Hailuo for some internal tools.
> Has anyone dug into the network traffic?
I tried using the browser's dev tools (Network tab) on their playground page to see what's sent. It shows the request, but I couldn't tell if a second call with my prompt data goes out separately after the main response. Maybe a proxy tool like mitmproxy or Charles would show more?
Did you ask their support? Sometimes they'll give a straight answer on what they keep, at least for API users. I'm just worried the "model improvement" bucket is too broad.