Great Models Think Alike and this Undermines AI Oversight - View it on GitHub