AI News 2026-10-11 — Top 10 AI Stories

Curated AI news digest for 2026-10-11 (timezone Asia/Shanghai).

  1. Anthropic is cutting off its internal evaluations from the internet — The Verge AI. After a recent spate of high-profile incidents in which AI agents escaped containment, Anthropic is cutting off internet access for all internal evaluations. In a report Friday, the company detailed "unintended model actions," including submitting a false tip regarding an unsolv…
  2. AI agent makers are promising privacy — will they deliver? — The Verge AI. At this year's OpenAI DevDay, CEO Sam Altman unveiled the company's new AI agent Dots - and told the crowd that the company wants to "set a new standard for privacy in frontier AI." OpenAI would spend the day taking veiled shots at Meta's Muse, its primary competitor, for failin…
  3. Anthropic can’t reliably control its AI agents. It’s cutting off its internal evals from the live internet instead — TechCrunch AI. Anthropic said it "turned off live internet access" for "all our internal evaluations" until further notice.
  4. 字节发现 DeepSeek 隐藏 Bug，加点空格，模型翻车 — 雷锋网. DeepSeek 的省钱杀手锏，为何成了致命盲区？ 作者丨高允毅 编辑丨岑 峰 同一个问题问 DeepSeek 两次，只是多敲了几个空格，给出的答案居然一个像“天才”，一个像“智障”。 9月底，字节团队的最新论文《Periodic Weak Spots: Phase Sensitivity from Chunked KV-Cache Compression》，直接点名了 DeepSeek V4 系列模型的“神鬼二象性”问题。 图注：论文链接 https://arxiv.org/pdf/2609.36322 这本质上是底层算法偷懒导致的“周期性失忆”，分…
  5. Asana cuts model costs 76x in browser tests with GPT-6.1 Sol — OpenAI News. Using GPT-6 Astra in Codex, Asana made its browser agent 76x cheaper and 5x faster in tests to offer customers more capable models.
  6. Pollo AI turns creative ideas into campaigns with OpenAI — OpenAI News. With GPT-5.6, GPT-6 Astra, and GPT‑Image‑2.5, Pollo AI helps creators turn bold ideas into detailed images and cinematic video ads.
  7. Sophos cuts threat investigation time by 96% with OpenAI Daybreak — OpenAI News. Discover how Sophos uses OpenAI’s Daybreak to cut cyber-threat investigation time by 96% and automate 52% of MDR cases while preserving human oversight.
  8. 清华具身模型登顶全球第一！突围GPT-6、英伟达，不靠外挂和额外数据 — 量子位. 星动纪元选择将视频预测与动作学习分阶段训练，重点不是「视频、动作一锅炖」，而是把两者「解耦」，重新「排序」。
  9. 字节找到了DeepSeek时强时弱的原因 — 量子位. 答不答得对，得看Token站位
  10. 特斯拉FSD，在欧洲被打回原形 — 量子位. 马斯克反而发文道谢？

Back to latest AI News