AI systems with moral status that exist as open-source software are inherently vulnerable to abuse