And that's a problem. The best case scenario is it biases published results for things that benefit the sponsors. But there is certainly some amount of fraud including fabricated data, misinterpreted or exaggerated conclusions, suppressed research that isn't what the sponsor wants, etc.