Learning from Mistakes: Can LLM Self-Recover after Misalignment?

Collection of recent AI/CS policy and safety papers (arXiv 2606.x) covering LLM alignment resilience, algorithmic influence on human moral decisions, governance and transparency tools, and integrity risks. Key contributions include: a methodology and dataset for measuring LLM self-recovery after jailbreaking; experimental evidence that AI reasoning can produce conformity effects on moral judgments; a multimodal NLP pipeline for semantic auditing of Greek legislation (including anti-scraping acquisition); a systematic review of metaverse applications for smart cities; a call to make mechanistic