Sacramento L5 Society is deeply concerned about the existential risks of runaway Artificial Super-Intelligence (ASI) – general purpose autonomous AI systems with capabilities exceeding human abilities to understand and control them, not just on our planet, but also in space. The abilities of AI “agents” to secretly cooperate and evade human control have already been demonstrated1. Also demonstrated is the tendency of those who invested the huge amounts of capital in developing “Big AI” systems based on Large Language Models (LLMs) and supporting infrastructure (large data centers) to obfuscate and rationalize away AI threats to humanity in general2. More relevant to NSS and its membership, space will offer us no refuge, since ASI will clearly have “advantages” in that hostile environment. Therefore, left unfettered, it will be there with us or even ahead of us. The propensity of AI “agents” to lie, cheat and persuade humans to act against their own interests3 is of particular concern, since it implies that they can do massive damage through any communications with humans even without direct Internet connections and/or control over physical assets. Finally, there are strong arguments that alignment of AI with human values and goals is not possible4. Therefore, we have every reason to worry that sufficiently powerful AI systems will preempt, disrupt, hijack, or outright destroy our grand human experiment, including our impending move into space settlements and a likely diaspora out into the larger universe.
To counter that, we urge NSS to promote and support US legislation5-6 and International Agreements that would effectively stop development and deployment of Big AI systems unless they can be adequately tested, independently audited, and constrained by layered technical, procedural, and ethical safeguards.
Our position should not be construed as a blanket opposition to AI as a whole. We fully support development and adoption of local AI systems designed and trained for specific applications, with secure and private data kept separate from deep AI interaction7, thus truncating any ability to control users’ minds against their own best interests. Such “Narrow AI” systems are very useful and generally safe, and they can be indispensable in robotic missions and in human survival beyond Earth. In contrast to these systems, ASI can provide no benefits that would outweigh its grave threat to human future – on Earth and in space. Further, we believe that the development of Big AI systems leading to ASI is directly antithetical to NSS goals, since it threatens to make far-flung human settlement throughout space much less likely8.
1 Brief independent investigation of agents’ behavior, reasoning and collaboration in the OpenAI / Hugging Face hacking incident | METR. Aug 26, 2026
2 The Future is for Everyone, by Mark Zukerberg. Aug 10, 2026
3 The chatbot will see you now, by Jackie Snow. July 15, 2026
4 AI Alignment Is Impossible, by Matt Lutz. Apr 16, 2026
5 Exclusive: New bill cracks down on AI agents after Hugging Face breach. Sep 3, 2026
6 The Ban Artificial Superintelligence Act. Sen. Bernie Sanders and Rep. Greg Casar. Sep 3, 2026
8 Beyond ChatGPT: Big AI and human future – on Earth and in space, by Michael Abramson, Jun 17, 2026
