2026-08-15 科技日报

扫描 330 篇候选内容 · 覆盖 650 个信息源 · 纳入 318 篇

📌 今日要点

• Woman claims her stepfather used Grok to transform childhood photo into explicit imagery
• Anthropic shares more details about how Claude’s new watermarks will work
• SpaceX officially closes its Cursor acquisition
• Investor pressure forces Nvidia to shrink its OpenAI bet just as Anthropic’s numbers defy bubble warnings
• AI-generated books are flooding Amazon and tanking sales for human authors
• Plaintiff hid invisible AI instructions in court filings to secretly influence automated review

📝 科技简讯

• Woman claims her stepfather used Grok to transform childhood photo into explicit imagery — TechCrunch AI
• Anthropic shares more details about how Claude’s new watermarks will work — TechCrunch AI
• SpaceX officially closes its Cursor acquisition — TechCrunch AI
• Investor pressure forces Nvidia to shrink its OpenAI bet just as Anthropic’s numbers defy bubble warnings — The Decoder
• AI-generated books are flooding Amazon and tanking sales for human authors — The Decoder
• Plaintiff hid invisible AI instructions in court filings to secretly influence automated review — The Decoder
• World Labs turns one real-world robot task into thousands of simulated variations for training — The Decoder
• The “tragedy of the cognitive commons” explains how rational AI adoption could destroy entire professions’ expertise — The Decoder
• New benchmark confirms AI models still perform poorly at visual perception — The Decoder
• 至知研究院提出大模型可解释性新路线：拆权重，数据成本不到 1% — 量子位
• 根治 AI 音乐通病！这家国产音乐模型正面挑战 SUNO — 量子位
• 源神启动！一张消费级显卡跑“Opus 级”Agent，Qwen3.8-27B 多项榜单反超 Claude — 量子位
• DeepSeek Harness 插件一夜燃爆 GitHub：长期记忆、电子宠物、4399 小游戏全来了 — 量子位
• 对话郎咸朋：用机器人创业重做一次“百万智驾量产” — 量子位
• 高效低成本开发，商业增长回报高，鸿蒙元服务背后的“开发厨房”让开发者按时吃饭！ — 量子位
• 刚刚，Anthropic 给 Claude 装了隐形卧底！全球生效 — 新智元
• 全球程序员都在给 Anthropic 白送钱！官方终于看不下去了 — 新智元
• OpenAI 研究员：100% 代码已交给 AI！Claude 重写 Python 库 — 新智元
• 刚刚，Cursor 彻底消失！ — 新智元
• Anthropic 自曝「私藏核武器」，Model 2 比 Mythos 5 更强！ — 新智元
• Position: Reasoning is a Learnable Rule-Based Process — arXiv AI (cs.AI)
• Diagnostic Foundation for Evaluating LLMs’ Research Integrity as Co-Scientists — arXiv AI (cs.AI)
• Position: The Alignment Community is Unintentionally Building a Censor’s Toolkit — arXiv AI (cs.AI)
• Agreement Is Not Alignment: Divergent Moral Grounds in Human and LLM Ethical Judgments — arXiv AI (cs.AI)
• Multi-Agent Scheduling with LLM-Assisted Contract Net Negotiation for Stream Processing in Mobile Edge Computing — arXiv AI (cs.AI)
• Kimi likes causal decision theory more after RL in twin prisoner’s dilemmas — LessWrong
• Luck is a function of surface area. — LessWrong
• What if Parameter Updates were Text? — LessWrong
• Using Chunked Monitoring to Detect Deception in Long Transcripts — LessWrong
• How To Catch a Distilled Model — LessWrong
• Hawaii braces for potential first direct hit by a hurricane in 34 years — BBC Science
• Firefighters battle wildfire tearing through Belgian nature reserve — BBC Science
• The influencers getting bitten while chasing the perfect shark shot — BBC Science
• Hollywood actor seeks help from vicar for series — BBC Science
• Cornish sardines back on menu for penguins — BBC Science
• Draft dark skies guidance drawn up for the Highlands — BBC Science
• Solar eclipse inspires talk of Pac-man and Shinto myths — BBC Science
• Secondhand book sales are booming. Is it because of AI? — BBC Tech
• Why tech bosses keep sharing their manifestos about AI — BBC Tech
• Memories Mysteriously Survive After Brain Shutdown, And Scientists Don’t Know Why — 404 Media

📊 数据概览