The contagion of fear
- ID
- 24377
- Status
- summarized
- Published
- 15 Sep 2026, 5:18 AM
- Fetched
- 15 Sep 2026, 5:49 AM
- Provider
- Simon Willison
- Category
- developer-ai
- Original URL
- https://simonwillison.net/2026/Sep/14/the-contagion-of-fear/
- Source URL
- https://simonwillison.net/atom/everything/
Summary
- Score
- 4.0
- Created
- 15 Sep 2026, 5:49 AM
- Tags
- Audience
- ai-ml-learnersai-agent-users
What happened
Bryan Cantrill pushes back against claims by former Anthropic employee Jacob Coxon that AI 'could kill us all by the end of the decade,' arguing that domain experts like Coxon lack expertise in critical infrastructure, bioweapons, or extinction biology, and are abusing public trust by making hand-wavy catastrophic predictions without elaboration. Cantrill draws on a personal lesson about causing unjustified panic among less technical peers and calls for circumspection when raising alarms.
Why it matters
This is an ethics-of-communication debate, not a technical or product change—builders don't need to alter any tooling or decisions. The practical takeaway is to be skeptical of catastrophic AI risk claims that lack domain-specific mechanistic detail, and to demand that anyone raising extinction-level alarms demonstrate relevant expertise in the domain they're speculating about (e.g., bioweapons, infrastructure).
Discussion angle
When an AI lab researcher makes an extinction-level claim, how do you evaluate whether their AI expertise transfers to the domain they're speculating about—and does this kind of discourse affect how you communicate AI capabilities and risks to non-technical stakeholders?