AI safety

Learning Resources

Light

• Writing Doom (27m video): short film on Superintelligence (2024)
• a casual intro to AI doom and alignment (2022)
• Anthony Aguirre: Keep The Future Human
  • Interactive Explainer
  • Essay: Keep the Future Human
  • We Can’t Stop AI – Here’s What To Do Instead (4m video, 2025)
  • The 4 Rules That Could Stop AI Before It’s Too Late (15m video, 2025)
• Tristan Harris TED talk (15m): Why AI is our ultimate test and greatest invitation
  • Text version: Center for Humane Technology: The Narrow Path: Why AI is Our Ultimate Test and Greatest Invitation
• Fable about Transformative AI

Deep

• The Compendium: Humanity risks extinction from its very creations — AIs. (2024)
• Introduction to AI Safety, Ethics, and Society (Dan Hendrycks, Center for AI Safety)
• AI Safety FAQ
• DeepMind short course on AGI safety

Description of Safety Concerns

Key Concepts

• Instrumental Convergence
• Orthogonality Thesis
• Inner/outer alignment
• Mesa-optimization
• Overhang
• Reward is not the optimization target (Alex Turner)

Medium-term Risks

• 2023-04: A.I. Dilemma – Tristan Harris and Aza Raskin” (video) (.website-files.com/5f0e1294f002b1bb26e1f304/64224a9051a6637c1b60162a_65-your-undivided-attention-The-AI-Dilemma-transcript.pdf podcast transcript): raises concern about human ability to handle these transformations
• 2023-04: Daniel Schmachtenberger and Liv Boeree (video): AI could accelerate perverse social dynamics
• 2023-10: Sociotechnical Safety Evaluation of Generative AI Systems (Google DeepMind)
• 2024-02: Towards a Cautious Scientist AI with Convergent Safety Bounds (Yoshua Bengio)
• 2024-07: Reasoning through arguments against taking AI safety seriously (Yoshua Bengio)
• 2025-04: AI-Enabled Coups: How a Small Group Could Use AI to Seize Power
• 2025-06: The Singapore Consensus on Global AI Safety Research Priorities

Long-term (x-risk)

• 2015-02: Sam Altman: Machine intelligence, part 1
• 2019-03: Daniel Kokotajlo and Wei Dai: The Main Sources of AI Risk?
• 2022-06: Eliezer Yudkowsky: List AGI Ruin: A List of Lethalities
• 2024-11: Marcus Arvan: ‘Interpretability’ and ‘alignment’ are fool’s errands: a proof that controlling misaligned large language models is the best anyone can hope for
• 2025-04: ASI existential risk: reconsidering alignment as a goal
• 2025-12: Philip Trammell and Leopold Aschenbrenner: Existential Risk and Growth

Status

• 2025-01: International Safety Report: The International Scientific Report on the Safety of Advanced AI (January 2025)
• AI Lab Watch (safety scorecard)

Assessmment

• AI Assessment Scale (AIAS): A practical framework to guide the appropriate and ethical use of generative AI in assessment design, empowering educators to make purposeful, evidence-based decisions
• 2025-07: Frontier AI Risk Management Framework in Practice: A Risk Analysis Technical Report

Policy

• 2024-07: On the Limitations of Compute Thresholds as a Governance Strategy Sara Hooker
• 2024-07: Framework Convention on Global AI Challenges (CIGI)
• 2024-08: NIST guidelines: Managing Misuse Risk for Dual-Use Foundation Models

Proposals

• 2025-02: Responsible AI Agents
• 2025-03: Control AI The Direct Institutional Plan
• 2025-04: Google DeepMind: Taking a responsible path to AGI
  • Paper: An Approach to Technical AGI Safety and Security

Research

• 2022-09: The alignment problem from a deep learning perspective
• 2022-12: Discovering Latent Knowledge in Language Models Without Supervision
• 2023-02: Pretraining Language Models with Human Preferences
• 2023-04: Do the Rewards Justify the Means? Measuring Trade-Offs Between Rewards and Ethical Behavior in the MACHIAVELLI Benchmark
• 2023-05: Model evaluation for extreme risks (DeepMind)
• 2023-05: Principle-Driven Self-Alignment of Language Models from Scratch with Minimal Human Supervision
• 2023-06: Preference Ranking Optimization for Human Alignment
• 2023-08: Self-Alignment with Instruction Backtranslation
• 2023-11: Debate Helps Supervise Unreliable Experts
• 2023-12: Weak-to-Strong Generalization: Eliciting Strong Capabilities with Weak Supervision (OpenAI, blog)
• 2023-12: Practices for Governing Agentic AI Systems (OpenAI, blog)
• 2024-01: Sleeper Agents: Training Deceptive LLMs that Persist through Safety Training (Anthropic)
• 2024-04: The Instruction Hierarchy: Training LLMs to Prioritize Privileged Instructions (OpenAI)
• 2024-07: On scalable oversight with weak LLMs judging strong LLMs
• 2024-07: Safetywashing: Do AI Safety Benchmarks Actually Measure Safety Progress? (Dan Hendrycks et al.)
• 2024-08: Tamper-Resistant Safeguards for Open-Weight LLMs (project, code)
• 2024-08: Better Alignment with Instruction Back-and-Forth Translation
• 2024-10: First-Person Fairness in Chatbots (OpenAI, blog)
• 2024-10: Sabotage evaluations for frontier models (Anthropic, blog)
• 2024-12: Alignment Faking in Large Language Models (Anthropic)
• 2024-12: Best-of-N Jailbreaking (code)
• 2024-12: Towards Safe and Honest AI Agents with Neural Self-Other Overlap
  • 2024-07: Self-Other Overlap: A Neglected Approach to AI Alignment
  • 2025-03: Reducing LLM deception at scale with self-other overlap fine-tuning
• 2024-12: Deliberative Alignment: Reasoning Enables Safer Language Models (OpenAI)
• 2025-01: Trading Inference-Time Compute for Adversarial Robustness (OpenAI, blog)
• 2025-01: Constitutional Classifiers: Defending against Universal Jailbreaks across Thousands of Hours of Red Teaming (Anthropic, blog,
• 2025-02: Utility Engineering: Analyzing and Controlling Emergent Value Systems in AIs (site, github)
• 2025-02: Auditing Prompt Caching in Language Model APIs
• 2025-03: The Alignment Problem from a Deep Learning Perspective
• 2025-03: Auditing language models for hidden objectives (Anthropic, blog)
• 2025-03: Superalignment with Dynamic Human Values
• 2025-04: Contemplative Wisdom for Superalignment
• 2025-04: Scaling Laws for Scalable Oversight (preprint, code)
• 2025-06: SHADE-Arena: Evaluating sabotage and monitoring in LLM agents (Anthropic, blog)
• 2025-06: Avoiding Obfuscation with Prover-Estimator Debate
• 2025-06: Persona Features Control Emergent Misalignment (OpenAI, blog)
• 2025-07: Why Do Some Language Models Fake Alignment While Others Don't? (Anthropic, code)
• 2025-07: Chain of Thought Monitorability: A New and Fragile Opportunity for AI Safety
• 2025-09: Detecting and reducing scheming in AI models
• 2025-11: Natural Emergent Misalignment from Reward Hacking in Production RL (Anthropic, blog)
• 2025-12: Difficulties with Evaluating a Deception Detector for AIs
• 2025-12: Monitoring Monitorability (OpenAI)
• 2026-01: Training large language models on narrow tasks can lead to broad misalignment
  • 2025-02: Preprint: Emergent Misalignment: Narrow finetuning can produce broadly misaligned LLMs

Demonstrations of Negative Use Capabilities

• 2024-12: Evaluating Large Language Models' Capability to Launch Fully Automated Spear Phishing Campaigns: Validated on Human Subjects
• 2025-04: Nathan Labenz (The Cognitive Revolution): AI Bad Behavior

Threat Vectors

• 2024-01: Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Training
• 2025-10: Poisoning Attacks on LLMs Require a Near-constant Number of Poison Samples

See Also

• AI predictions