He has a documented history (charitably) overly-optimistic claims and (accurately) lies. On multiple occasions, he (or an unnamed disgruntled employee who happens to share his exact views) has manipulated the output of his own chatbot to push particular political positions.
He's very unlikely to actually do this, and if he did, it would not do what he describes.
The incentives are completely backwards in his case, and his history shows a blatant disregard for accuracy or responsible use of information.