As AI systems become more capable, the risks they pose potentially increase. Responsible scaling policies - sometimes called "responsible scaling commitments" or "preparedness frameworks" - are commitments by AI developers to assess risks before training or deploying more capable models and to pause or modify development if certain risk thresholds are crossed. Anthropic's Responsible Scaling Policy, OpenAI's Preparedness Framework, and Google DeepMind's Frontier Safety Framework are prominent examples. These policies typically define capability levels, associated risks, and the safety measures required at each level. The key question is whether these commitments are meaningful and enforceable. Critics point out that they're voluntary, that the companies themselves define the risk thresholds, and that commercial pressure to deploy new models creates a conflict of interest. Proponents argue that even imperfect self-regulation is better than none, and that it buys time for external governance to catch up. Some governments are beginning to build on these voluntary commitments, exploring requirements for pre-deployment testing of frontier models and mandatory reporting of certain capabilities. For businesses using or building on frontier models, understanding your provider's scaling policy helps you assess the stability of the platform you're depending on and the seriousness of their safety commitments.