Anthropic 发布 Claude Mythos Preview 对齐风险报告

information to a degree not justified by any commensurate public benefit. Redacted text has been made available to Anthropic’s internal Stress-Testing Team—alongside technical employees of Anthropic more broadly—and will be made available to some external reviewers. All such reviewers, internal and external, will be asked to comment on the appropriateness of the redactions. 4 2 Overview Threat model An AI model with access to powerful affordances within an organization could use its affordances [...] and further reduce the rate of concerning behavior occurring during training. Overall risk assessment Very low, but higher than for previous models. 5 3 Threat model The threat: If AI models are heavily used by powerful organizations—for example, to write a large amount of critical code—they might take advantage of this access to adversely influence decision-making, insert and exploit cybersecurity vulnerabilities, and take other actions that could significantly raise the risk of future harmful [...] use its affordances to autonomously exploit, manipulate, or tamper with that organization’s systems or decision-making in a way that raises the risk of future significantly harmful outcomes (e.g. by altering the results of AI safety research). Relevant AI model We focus our analysis on Claude Mythos Preview, our latest, most capable model. Current usage, capabilities and propensities Claude Mythos Preview is used heavily within Anthropic for coding, data generation, and other agentic use cases.

原文版权归 官方博客 · Anthropic / Meta / xAI 及原作者所有，本站转载仅供阅读。如需删除，请联系 support@aisa.one。