Large language models (LLMs) can write code in "hardware languages" defining circuits. To learn if LLMs can be asked to add vulnerabilities to hardware, we ran a global competition. We found that teams could introduce even severe vulnerabilities, implying insufficient LLM guardrails.
No takes yet. Share an insight, caveat, or question.
Blocklove et al. (2025) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: