Every so often a company says the quiet part into a microphone. A Twitch executive explained that its parent company chose to use streamers' content for AI training by default — because, in the exec's words, “nobody would opt in.”
Sit with that logic for a second. The platform knew creators would say no if asked. So the ask was removed. Thousands of hours of streams — voices, faces, personalities, the actual product creators sell — became training fuel unless someone found the toggle and flipped it off. This is the new default across the industry: your content feeds the machine until you actively object.
The rules are starting to push back. California's training-data transparency law, in force since January, requires AI developers to disclose what went into their models, and more states are circling the same territory. But regulation moves in years; platform settings change in quiet Tuesday updates.
So the move is on you, and it's simple. Once a quarter, open the settings page of every platform that hosts your work and search for “AI,” “machine learning” or “training.” Read the clause before you upload the thing you can't afford to give away. And if you build software: consent tooling, provenance tracking and opt-out management just became a product category with guaranteed demand — because as this executive helpfully confirmed, the platforms will not build it for you.
