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。
The Decoder这条资讯聚焦“AI 研究与论文进展”：Researchers may have found a way to stop AI models from intentionally playing dumb during safety evaluations。原始摘要提到：A study by researchers from the MATS program, Redwood Research, the University of Oxfo…建议研究人员、算法团队、技术内容作者和前沿观察者重点关注它可能带来的工具入口、工作流、成本、风险或选型变化；原文链接已保留，便于继续阅读完整报道。
The Decoder这条资讯聚焦“模型发布与实时多模态能力”：GPT-5.5 costs 49 to 92 percent more than its predecessor, depending on the input length。原始摘要提到：OpenAI doubled GPT-5.5's list price compared to GPT-5.4, claiming shorter responses wo…建议模型使用者、产品经理、内容团队和 API 接入团队重点关注它可能带来的工具入口、工作流、成本、风险或选型变化；原文链接已保留，便于继续阅读完整报道。
The Decoder这条资讯聚焦“模型发布与实时多模态能力”：METR says it can barely measure Claude Mythos, Palo Alto Networks warns of autonomous AI attackers。原始摘要提到：METR can barely measure Claude Mythos Preview with its current test suite. Only five o…建议模型使用者、产品经理、内容团队和 API 接入团队重点关注它可能带来的工具入口、工作流、成本、风险或选型变化；原文链接已保留，便于继续阅读完整报道。
The Decoder这条资讯聚焦“全球 AI 产业动态”：ByteDance plans over $30 billion for AI expansion, bets big on Chinese chips。原始摘要提到：ByteDance is raising its planned AI spending for 2026 to over 200 billion yuan (roughl…建议关注全球 AI 产品、公司、工具和趋势变化的读者重点关注它可能带来的工具入口、工作流、成本、风险或选型变化；原文链接已保留，便于继续阅读完整报道。
The Decoder这条资讯聚焦“全球 AI 产业动态”：Anthropic and OpenAI sit down with religious leaders to seek ethical advice。原始摘要提到：Anthropic and OpenAI are turning to religious leaders for help with AI ethics. At the …建议关注全球 AI 产品、公司、工具和趋势变化的读者重点关注它可能带来的工具入口、工作流、成本、风险或选型变化；原文链接已保留，便于继续阅读完整报道。
The Decoder这条资讯聚焦“AI 研究与论文进展”：AI agents can now hack computers and copy themselves, and they're getting better fast。原始摘要提到：Palisade Research shows that AI agents can hack remote computers, copy themselves onto…建议研究人员、算法团队、技术内容作者和前沿观察者重点关注它可能带来的工具入口、工作流、成本、风险或选型变化；原文链接已保留，便于继续阅读完整报道。
The Decoder这条资讯聚焦“全球 AI 产业动态”：Broadcom reportedly won't build OpenAI's custom chip unless Microsoft buys 40 percent of them。原始摘要提到：OpenAI's custom AI chip project with Broadcom has hit a funding wall. Broadcom won't f…建议关注全球 AI 产品、公司、工具和趋势变化的读者重点关注它可能带来的工具入口、工作流、成本、风险或选型变化；原文链接已保留，便于继续阅读完整报道。
The Decoder这条资讯聚焦“模型发布与实时多模态能力”：Fields Medalist says ChatGPT 5.5 Pro delivered "PhD-level" math research in under two hours with zero human help。原始摘要提到：Fields Medalist Timothy Gowers had ChatGPT 5.5 Pro tackle open problems in number theo…建议模型使用者、产品经理、内容团队和 API 接入团队重点关注它可能带来的工具入口、工作流、成本、风险或选型变化；原文链接已保留，便于继续阅读完整报道。