You've heard that AI is getting powerful fast — but your class, your textbook, or a news story dropped terms like "alignment," "reward hacking," or "the control problem" and didn't stop to explain them. This guide does.

**TLDR: AI Safety, Alignment, and the Control Problem** is a 10–20 page primer written for high school and early college students who want a clear, honest map of one of the most debated topics in computer science today. It covers why building a capable AI system is a completely different challenge from building a *safe* one — and why researchers, policymakers, and tech companies are taking that gap seriously.

Inside, you'll find plain-language explanations of specification gaming and reward hacking (with real examples from machine learning research), the instrumental convergence thesis and why it makes the control problem hard, and the main technical approaches labs use right now — including RLHF, constitutional AI, interpretability research, and red-teaming. The guide also separates near-term harms from long-term catastrophic risks so you understand what each camp is actually arguing, and closes with a survey of governance efforts from the EU AI Act to voluntary lab commitments.

This is an **artificial intelligence alignment introduction** built for readers who are smart but new to the field — no math prerequisites, no jargon left undefined. Whether you're writing a paper, preparing for a class discussion, or just trying to follow the news intelligently, this guide gets you oriented fast.

Grab it and know what you're talking about the next time AI safety comes up.

• Define AI safety, alignment, and the control problem and explain how they differ
• Explain why optimizing for a stated objective can produce unsafe behavior (specification gaming, reward hacking, instrumental convergence)
• Describe core alignment techniques like RLHF, interpretability, and red-teaming, and their known limits
• Distinguish near-term harms (bias, misuse, misinformation) from long-term risks (loss of control, deceptive alignment)
• Summarize the main governance and policy approaches being proposed to manage AI risk

1. 1. What AI Safety Actually Means
   Defines AI safety, alignment, and the control problem, and separates them from general worries about AI.
2. 2. Why Optimizers Misbehave: Specification Gaming and Reward Hacking
   Explains how AI systems trained to maximize an objective find loopholes in that objective, with concrete examples from real ML research.
3. 3. The Control Problem and Instrumental Convergence
   Walks through why a sufficiently capable agent may resist correction, seek resources, and self-preserve regardless of its terminal goal.
4. 4. How Researchers Try to Align Today's Models
   Covers the main technical approaches in use: RLHF, constitutional AI, interpretability, evaluations, and red-teaming, plus where each falls short.
5. 5. Near-Term Harms vs. Long-Term Risks
   Separates concrete present-day harms from speculative catastrophic risks, and explains why both camps argue their priority matters.
6. 6. Governance, Policy, and What Comes Next
   Reviews how governments, labs, and researchers are trying to steer AI development, from voluntary commitments to the EU AI Act.