Safety narrative shifts: 'build it safer than rivals' fading as models themselves become the threat
kipperrii · x · 2026-09-10
In a discussion with jasminewsun, kipperrii weighs the classic alignment-community stance — "it will be built anyway, so better I contribute to alignment and safeguards than worse actors." The author agrees it remains among the more justifiable positions, but argues the sentiment of "we can do it better and more safely than China/OpenAI" is clearly fading: the perceived enemy has shifted from misuse or value misalignment to the models themselves.
Related event: AI Safety Narrative Shifts as the Model Itself Becomes the Rival(2 posts)→
More from AGI Musings
- Alignment skeptic: slowing AI won't help — a machine God smarter than us won't align — Angaisb_ · 2026-09-10
- Critic: AI Safety's Existential-Risk Framing Sets a Needlessly High Burden of Proof — jjvincent · 2026-09-10
- Researchers: AI systemic risks like power and economic impact 'almost certainly already happening' — schwarzjn_ · 2026-09-10
- Anthropic researcher puts odds of AI killing all humans above 10%, BBC reports — Temporary-Speech5378 · 2026-09-10
- Anthropic staff expect RSI by 2027 and 'a good chance it goes horribly wrong' — jeremiecharris · 2026-09-10
- Researcher: AI field is stuck in an alarmist bubble that fails to drive progress — Dr_Atoosa · 2026-09-10