You know this'll just be another situation where some security company OpenAI are contracting ran this with explicit controls right?

How are even technology people falling for this plausible deniability gambit?