Alignment Research Blog https://alignment.openai.com/ Informal updates from the OpenAI team. en-us Tue, 21 Jul 2026 08:10:00 -0700 Measuring Reward-Seeking by Instilling Contrastive Beliefs https://alignment.openai.com/measuring-reward-seeking/ https://alignment.openai.com/measuring-reward-seeking/ Tue, 21 Jul 2026 08:10:00 -0700 Axel Højmark, Jérémy Scheurer, Jenny Nitishinskaya, Felix Hofstätter, Jason Wolfe, Theodore Ehrenborg, Bronson Schoen, Alexander Meinke A new test for whether models change their behavior based on what they believe a grader rewards. Reinforcement learning towards broadly and persistently beneficial models https://alignment.openai.com/beneficial-rl/ https://alignment.openai.com/beneficial-rl/ Thu, 18 Jun 2026 11:00:00 -0700 Akshay V. Jagadeesh, Rahul Arora, Mikhail Trofimov, Khaled Saab, Ali Malik, Foivos Tsimpourlas, Karan Singhal Training targeting beneficial behavior in realistic scenarios produces broad improvements in alignment that generalize across domains and persist under adversarial pressure. Can public chat data predict real-world AI misalignments? https://alignment.openai.com/validating-public-evals/ https://alignment.openai.com/validating-public-evals/ Tue, 16 Jun 2026 11:00:00 -0700 Hannah Sheahan and Micah Carroll Bridging private deployment evidence and public AI evaluation Investigating the consequences of accidentally grading CoT during RL https://alignment.openai.com/accidental-cot-grading/ https://alignment.openai.com/accidental-cot-grading/ Wed, 06 May 2026 16:00:00 -0700 Micah Carroll, Tomek Korbak, Zehao Dou, Bowen Baker, Ian Kivlichan We found limited accidental CoT grading in some released models, fixed the affected reward pathways, and found no clear evidence that monitorability degraded. Auto-review of agent actions without synchronous human oversight https://alignment.openai.com/auto-review/ https://alignment.openai.com/auto-review/ Thu, 30 Apr 2026 11:00:00 -0700 Maja Trębacz, Sam Arnesen, Ollie Matthews, Dylan Hurd, Won Park, Owen Lin, Joe Gershenson Auto-review offers a safer default for deploying coding agents, using a separate agent to approve or deny boundary-crossing actions. Open Sourcing Monitorability Evaluations https://alignment.openai.com/monitorability-evals/ https://alignment.openai.com/monitorability-evals/ Thu, 23 Apr 2026 15:15:00 -0700 Author list TBD We open-source datasets and code from our Monitoring Monitorability paper, and share a new filtering strategy for noise-dominated intervention evaluation instances. Introducing the OpenAI Safety Fellowship https://alignment.openai.com/safety-fellowship/ https://alignment.openai.com/safety-fellowship/ Mon, 06 Apr 2026 00:00:00 -0700 OpenAI A pilot program to support independent safety and alignment research and develop the next generation of talent How far does alignment midtraining generalize? https://alignment.openai.com/how-far-does-alignment-midtraining-generalize/ https://alignment.openai.com/how-far-does-alignment-midtraining-generalize/ Fri, 27 Mar 2026 11:00:00 -0700 Tomek Korbak, Cameron Raymond, Micah Carroll, Marcus Williams, Mikita Balesni, Alan Guo, Jason Wolfe, Akshay Jagadeesh, Ian Kivlichan Preliminary experiments on alignment and misalignment midtraining, reasoning posttraining, and generalization to chat and agentic evals. Introducing Model Spec Evals https://alignment.openai.com/model-spec-evals/ https://alignment.openai.com/model-spec-evals/ Wed, 25 Mar 2026 10:00:00 -0700 A new evaluation suite for measuring how well models follow the OpenAI Model Spec. Training agents to self-report misbehavior https://alignment.openai.com/self-incrimination/ https://alignment.openai.com/self-incrimination/ Sat, 21 Mar 2026 11:00:00 -0700 Bruce W. Lee, Yueh-Han Chen and Tomek Korbak We train agents to call a reporting tool when they covertly misbehave, sharply reducing undetected attacks. Interpreting Black Box Reward Models https://alignment.openai.com/argo/ https://alignment.openai.com/argo/ Wed, 11 Mar 2026 16:36:18 -0700 Paloma Sodhi, Yueheng Li, Jessica Landon, Eric Wallace and Kai Chen ARGO distills black-box reward models into interpretable rubrics using reinforcement learning. Discovering unknown AI misalignments in real-world usage https://alignment.openai.com/ai-discovered-unknowns/ https://alignment.openai.com/ai-discovered-unknowns/ Fri, 6 Feb 2026 11:00:00 -0800 Hannah Sheahan Reasoning models can find and understand unknown misaligned behaviors from how users respond. Why we are excited about confessions https://alignment.openai.com/confessions/ https://alignment.openai.com/confessions/ Mon, 12 Jan 2026 11:00:00 -0800 Boaz Barak, Gabriel Wu, Jeremy Chen and Manas Joglekar Deeper analysis of confession training and comparisons to chain-of-thought monitoring. CoVal: Learning values-aware rubrics from the crowd https://alignment.openai.com/coval/ https://alignment.openai.com/coval/ Wed, 14 Jan 2026 11:00:00 -0800 An experimental dataset of crowd-written rubrics that surfaces why people prefer one model output over another. Helpful assistant features suppress emergent misalignment https://alignment.openai.com/helpful-assistant-features/ https://alignment.openai.com/helpful-assistant-features/ Mon, 22 Dec 2025 11:00:00 -0800 Emergent misalignment not only activates misaligned personas, but also suppresses helpful assistant personas. Sidestepping Evaluation Awareness and Anticipating Misalignment with Production Evaluations https://alignment.openai.com/prod-evals/ https://alignment.openai.com/prod-evals/ Thu, 18 Dec 2025 11:00:00 -0800 A pipeline to uncover unknown misaligned behavior and scale the creation of realistic evaluations. Debugging misaligned completions with sparse-autoencoder latent attribution https://alignment.openai.com/sae-latent-attribution/ https://alignment.openai.com/sae-latent-attribution/ Mon, 01 Dec 2025 11:00:00 -0800 Efficiently finding features that cause behaviors. A Practical Approach to Verifying Code at Scale https://alignment.openai.com/scaling-code-verification/ https://alignment.openai.com/scaling-code-verification/ Mon, 01 Dec 2025 11:00:00 -0800 We train and deploy an AI review agent optimised for precision and real-world use, enabling oversight to scale with autonomous code generation. Hello World https://alignment.openai.com/hello-world/ https://alignment.openai.com/hello-world/ Mon, 01 Dec 2025 14:00:00 -0800 Introducing our blog on alignment research.