Sam Altman’s three AI safety scenarios range from global oversight to a worst-case outcome: Report Firstpost