All Articles
1 min read
Исследования2026年10月22日
Behavior Beyond Text: Evaluating Refusal Transfer Failure in LLM Agents
arXiv:2610.28982v1 Announce Type: new Abstract: While large language models are increasingly deployed as tool-calling agents, their safety evaluations still focus primarily on their ability to refuse harmful requests in text. We investigate whether the refusal behavior in text…
Ekagra Gupta, Raj Saha, Aniekan-Abasi Inyang, Adam Al Gharib·arXiv — cs.AI
Original SourceKey Takeaways
- 01.arXiv:2610.28982v1 Announce Type: new Abstract: While large language models are increasingly deployed as tool-calling agents, their safety evaluations still focus primarily on their ability to refuse harmful requests in text
- 02.We investigate whether the refusal behavior in text…
This material is provided as an extended analytical summary.
Read full article at sourceAggregated within AI Qazaqstan national intelligence stream with author attribution.
Related Intelligence
All stories →Исследования1 мин
GPT-6.1 Terra：OpenAI 将中端代理模型的价格减半
🖥️ GPT-6.1 Terra：OpenAI 将中档代理模型的价格减半 OpenAI 发布了 GPT-6.1 Terra，用于基于代理的场景、计算机工作和日常编码。 API 价格是 GPT-5.6 Terra 的一半：每百万输入代币 1 美元，而不是每百万输入代币 2 美元和 6 美元……
Telegram — AI Machine Learning Big Data
Исследования1 мин
GPT-6.1 Terra 现已进入轨道
GPT-6.1 Terra：以较低的价格提供快速且功能强大的计算机使用。 GPT-6.1 Terra 可以理解屏幕上的内容，跨应用程序执行操作，自然书写，并且是适合您整个工作日的更强大的模型，可以高效地处理更多工作。 GPT-6.1 Terra 继续……
YouTube — OpenAI Official
Исследования1 мин
真正的人类请站起来吗
这个故事最初出现在我们关于人工智能的每周通讯《算法》中。要首先在您的收件箱中收到此类故事，请在此处注册。经过数周的疯狂猜测，Anthropic 已提交 IPO 招股说明书。这份 335 页的文件（大约 100 页是关于风险的）……
MIT Technology Review