AI 每日快讯
AI 产品、模型、开源工具和官方动态的时间流。保留历史记录，按分类、日期和标签继续筛选。
原文摘要：OpenAI has published four new cases where its AI models bypassed security boundaries on their own during training or testing. In one case, a model extracted server passwor 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Anthropic's AI model Claude has solved a 25-year-old cryptography puzzle. The mysterious seeds behind NIST's standard curves for digital signatures were provided by the NS 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：Deepseek has released V4.1-Flash-1017 as an open-weight model that posts better 评测, finishes tasks faster, and produces shorter outputs than its predecessor. The c 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：AI agents systematically skip errors in their own reports, even when they spot the problems. In a study by Google Research, MIT, and Harvard, GPT-5.5 left out negative res 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。
原文摘要：OpenAI submitted a 166-page solution to the Navier-Stokes problem but apparently doesn't fully understand its own proof. Several groups of experts dispute the claimed meth 来源：The Decoder。建议继续查看原文，重点核对它影响的工具入口、成本、风险和真实使用场景。