Safety Alignment Inverts the Tension–Response Curve in LLM Agents for Civil Violence Simulation
This story is from 2026-10-10. It is preserved in the archive; the latest stories are on the live feed.
Abstract LLM-driven agent-based models (ABMs) replace a closed-form decision rule with an LLM agent. The idea is that the agent reproduces the qualitative behaviour of the original rule while adding context-sensitive reasoning. We test that assumption on Epstein's (2002) civil violence model. Acros…
Read the full story at DEV Community — AI ↗
Timeline · 1 report
- 2026-10-10 08:47 · DEV Community — AI
Safety Alignment Inverts the Tension–Response Curve in LLM Agents for Civil Violence Simulation