The ethical bar isn't "never do human testing on anybody ever".
If an A/B test is likely to cause people harm (the 2012 Facebook Depression Experiment comes to mind), then it should absolutely be held to the same IRB standard as academia.
The point of IRB is to ensure that your human testing doesn't cause anybody any material or psychological harm. "Human testing" isn't a bad word, it's just something that needs to be done with care.
If an A/B test is likely to cause people harm (the 2012 Facebook Depression Experiment comes to mind), then it should absolutely be held to the same IRB standard as academia.
The point of IRB is to ensure that your human testing doesn't cause anybody any material or psychological harm. "Human testing" isn't a bad word, it's just something that needs to be done with care.