AI safety advocates: it's time to be more self-reflective about our personal usage of frontier AI
connorsscratchpad.substack.com · 2,775 words · saved by 1 readers
Match your spending to your values, avoid cognitive dissonance, and protect your moral commitments from getting hijacked.
The AI safety community has to this point been quite nonchalant about our own personal use of AI. The common view has been that it’s only future AI models that pose real dangers, so it doesn’t matter what we do now with currently existing models. For a time, this was reasonable: existing models were essentially deeply flawed toys, AI companies were years from IPO, and the pace of progress was slow compared to today’s dizzying rate of advance. Now, in the summer of 2026, the situation has changed. Our use of advanced AI models has far more of an impact than it does in the past, and poses new…
saved by
related reading
- A History of the Future, 2025-2027 - by Rudolf Lainenosetgauge.com
- Core views on AI safety: When, why, what, and how \ Anthropicanthropic.com
- Dario Amodei — We Must Pace the Frontierdarioamodei.com
- The Artificiality of Alignmentjoinreboot.org
- AI in 2025: gestalt — LessWronglesswrong.com
- Leaving Open Philanthropy, going to Anthropic - Joe Carlsmithjoecarlsmith.com
- Why I left Anthropic’s safety team to hold AI companies accountablejbenton1.substack.com
- My picture of the present in AI — LessWronglesswrong.com
- A guide to the AI tribes - by Michel Justen - What is thismicheljusten.substack.com
- AI safety is not a model propertysubstack.com
- Pacing the Frontierpacingthefrontier.com
- Common Elements of Frontier AI Safety Policiesmetr.org