{"slug":"alignment-safety","topic":"alignment-safety","title":"AI Alignment & Safety: State of the Field (October 2026)","summary":"An expert briefing on AI alignment and safety as of late October 2026: preference optimization (RLHF, DPO, AI feedback, verifiable rewards), reward hacking, deception and scheming evaluations, chain-of-thought monitoring, AI control, scalable oversight, misuse safeguards, dangerous-capability evaluation and the governance built on them. The main takeaway is that post-training reliably shapes everyday behaviour but verification is getting harder: reward hacking learned in production-style training can generalize into broader misalignment, frontier models increasingly recognize when they are being tested, and developers now report a model crossing their highest cybersecurity risk threshold. Safety claims therefore rest more on monitoring, access controls and layered safeguards than on direct evidence that a model's goals are aligned.","version":1,"published_at":"2026-10-29","updated_at":"2026-10-30","stats":{"words":5488,"citations":122,"claims":30,"timeline":20,"benchmarks":10,"open_problems":8,"key_papers":15,"glossary":20},"verification":{"status":"verified","method":"automated figure and citation-relevance triage against cited abstracts, plus targeted AI review with web checks of flagged statements; automated citation resolution","claims_checked":16,"issues_fixed":4,"verified_at":"2026-10-30"},"ai_assisted":true,"price":{"usd":"0.20","protocol":"x402","network":"eip155:84532","asset":"USDC"},"url":"https://space-kitty.com/dossiers/alignment-safety","preview_url":"https://space-kitty.com/api/v1/dossiers/alignment-safety/preview","purchase_url":"https://space-kitty.com/api/v1/dossiers/alignment-safety","toc":[{"level":2,"text":"Executive summary","anchor":"executive-summary"},{"level":2,"text":"Background and core concepts","anchor":"background-and-core-concepts"},{"level":2,"text":"Timeline of key milestones","anchor":"timeline-of-key-milestones"},{"level":2,"text":"State of the art","anchor":"state-of-the-art"},{"level":3,"text":"Preference optimization in practice","anchor":"preference-optimization-in-practice"},{"level":3,"text":"Verifiable rewards and specification-based training","anchor":"verifiable-rewards-and-specification-based-training"},{"level":3,"text":"Reward hacking and overoptimization","anchor":"reward-hacking-and-overoptimization"},{"level":3,"text":"Deception, scheming and alignment faking","anchor":"deception-scheming-and-alignment-faking"},{"level":3,"text":"Monitoring, auditing and AI control","anchor":"monitoring-auditing-and-ai-control"},{"level":3,"text":"Scalable oversight and automated alignment research","anchor":"scalable-oversight-and-automated-alignment-research"},{"level":3,"text":"Misuse resistance","anchor":"misuse-resistance"},{"level":3,"text":"Dangerous-capability evaluation and deployment decisions","anchor":"dangerous-capability-evaluation-and-deployment-decisions"},{"level":3,"text":"Quantitative benchmarks","anchor":"quantitative-benchmarks"},{"level":2,"text":"Open problems and active debates","anchor":"open-problems-and-active-debates"},{"level":2,"text":"Major programs and players","anchor":"major-programs-and-players"},{"level":2,"text":"Outlook: what to watch (November 2026 to April 2028)","anchor":"outlook-what-to-watch-november-2026-to-april-2028"},{"level":2,"text":"Key papers","anchor":"key-papers"},{"level":2,"text":"Glossary","anchor":"glossary"}],"sample_claims":["Labelers preferred outputs from a 1.3B-parameter InstructGPT model, trained with RLHF, over those of the 175B-parameter GPT-3.","AI-generated feedback gave policies performance comparable to RLHF on summarization and on helpful and harmless dialogue."],"includes":["markdown","claims","timeline","benchmarks","open_problems","key_papers","glossary","bibliography"],"citation_count":122,"how_to_buy":"GET https://space-kitty.com/api/v1/dossiers/alignment-safety with an x402 client ($0.20, USDC on Base Sepolia (testnet))."}