Built to Learn, Trained to Harm: A Comprehensive Survey of Artificial Intelligence and Ethics

Artificial intelligence has gone from a niche academic curiosity to something that affects nearly every part of daily life, and it has done so faster than most people were ready for. This survey report is an attempt to put together, in plain and honest terms, a wide-ranging overview of what AI actually is, how it works under the hood, how it shows up in the platforms people use every day, and where things start going wrong from an ethical standpoint. The topics covered here span the technical and the moral: from the basic definitions of narrow and general AI, to the mechanics of large language models and generative systems, through to harder questions about fairness, privacy, transparency, accountability, trust, and the misuse of AI to create harmful content including child sexual abuse material. This is not a paper that pretends AI is all good or all bad. The research surveyed here, drawn from publications between 2020 and early 2026, reflects a field genuinely wrestling with what it has built. Real incidents are discussed throughout, including the Grok undressing scandal and documented cases of AI-generated child sexual abuse material, because sanitizing those incidents out of a report on AI ethics would defeat the whole point. The goal of this survey is to give anyone who reads it, whether they are a student, a policymaker, or just a curious person, a solid, citation-backed foundation for understanding where AI stands, what the research says about its problems, and why those problems matter.