Why Hard AI Takeover Scenarios Are Unlikely: A Concrete Skeptical Breakdown
sebkrier · x · 2026-11-21
Sebkrier endorses Jake Steckler's new Substack essay pushing back on the AI safety community's belief that rogue AI could seize power militarily. The piece argues that while sandbox escapes and hacking warrant concern, no one persuasively explains how compromised servers or bribed humans could physically defeat a modern military. Steckler instead urges focusing on near-term risks from poorly controlled AI in military settings, whose solutions also mitigate catastrophic scenarios and are an easier sell to policymakers.
More from AGI Musings
- "Superintelligence Becomes a Commodity": Why the Model Layer Has No Moat — markjeffrey · 2026-11-21
- OpenAI researcher re-ups 5-year-old bet: AI to crack 6 of 7 math problems by 2040 — littmath · 2026-11-21
- Proposal: 500 randomly picked Americans would draft AI policy that becomes law unless Congress blocks it in a month — edelwax · 2026-11-21
- Markets aren't pricing in AGI: efficient-market believers and short-timelines fans can't both be right — jessi_cata · 2026-11-21
- AI21's Yoav Goldberg calls for strong 'open-thought' models with fully transparent CoT — ChrSzegedy · 2026-11-21
- Atari E.T. creator likens AI-made game flood to the glut before the 1983 video game crash — ThomasSimonini · 2026-11-21