If the risks OpenAI and its principals use as a rationale for not being even slightly open were anywhere close to as significant as they paint them, it would be a reason for either
intense transparency, public scrutiny, and public control of AI or, if even that were deemed too dangerous, public prohibition and ruthless, even violent suppression of AI research.
What it would most emphatically not be is a rationale for it to be tightly controlled by large for-profit corporations, who are extremely bad at and structurally disincentivized from responsibly managing external risks.