Goto

Collaborating Authors

 contact


How Brittle is Agent Safety? Rethinking Agent Risk under Intent Concealment and Task Complexity

Ma, Zihan, Zhu, Dongsheng, Liu, Shudong, Zhang, Taolin, Liu, Junnan, Li, Qingqiu, Luo, Minnan, Zhang, Songyang, Chen, Kai

arXiv.org Artificial Intelligence

Current safety evaluations for LLM-driven agents primarily focus on atomic harms, failing to address sophisticated threats where malicious intent is concealed or diluted within complex tasks. We address this gap with a two-dimensional analysis of agent safety brittleness under the orthogonal pressures of intent concealment and task complexity. To enable this, we introduce OASIS (Orthogonal Agent Safety Inquiry Suite), a hierarchical benchmark with fine-grained annotations and a high-fidelity simulation sandbox. Our findings reveal two critical phenomena: safety alignment degrades sharply and predictably as intent becomes obscured, and a "Complexity Paradox" emerges, where agents seem safer on harder tasks only due to capability limitations. By releasing OASIS and its simulation environment, we provide a principled foundation for probing and strengthening agent safety in these overlooked dimensions.


Is Russia using my childhood home as a military base?

BBC News

Use BBC.com or the new BBC App to listen to BBC podcasts, Radio 4 and the World Service outside the UK. Is Russia using my childhood home as a military base? Vitaly's home village of Verkhnya Krynytsya in the Zaporizhzhia region was occupied by Russia shortly after the start of the full-scale invasion in February 2022. Now, in a Ukrainecast exclusive, he tells Victoria why it's likely his childhood home is being used as a base by the Russian military. Plus, BBC Verify has revealed a surge in Ukrainian drone strikes on Russian oil refineries in recent months.


Here's what I made of Snap's new augmented-reality Spectacles

MIT Technology Review

These fifth-generation Spectacles can display visual information and applications directly on their see-through lenses, making objects appear as if they are in the real world. The interface is powered by the company's new operating system, Snap OS. There is no screen covering your field of view. Instead, images appear to float and exist in three dimensions in the world around you, hovering in the air or resting on tables and floors. Snap CTO Bobby Murphy described the intended result to MIT Technology Review as "computing overlaid on the world that enhances our experience of the people in the places that are around us, rather than isolating us or taking us out of that experience."





How Artificial Intelligence (AI) will impact Hollywood

#artificialintelligence

AI is rapidly changing the way Hollywood functions. It revolutionizes how stories are told, how movies are made, how audiences engage with content, and more. AI has the potential to disrupt the entire movie industry, from the way producers develop scripts to the way audiences consume content. AI is already being used to help filmmakers create more engaging stories. AI-powered screenwriting tools are being used to help writers generate ideas and structure their stories.


GitHub - unitaryai/detoxify: Trained models & code to predict toxic comments on all 3 Jigsaw Toxic Comment Challenges. Built using Pytorch Lightning and 🤗 Transformers. For access to our API, please email us at contact@unitary.ai.

#artificialintelligence

Trained models & code to predict toxic comments on all 3 Jigsaw Toxic Comment Challenges. Built using ⚡ Pytorch Lightning and 🤗 Transformers. For access to our API, please email us at contact@unitary.ai. - GitHub - unitaryai/detoxify: Trained models & code to predict toxic comments on all 3 Jigsaw Toxic Comment Challenges. Built using ⚡ Pytorch Lightning and 🤗 Transformers. For access to our API, please email us at contact@unitary.ai.


Ever AI

#artificialintelligence

No code full stack data science platform - data warehousing, data preparation, data modeling, data visualization, machine learning model deployment, training and validation in one place.

  Country: Asia > Malaysia (0.10)

Digital Era of Pharmaceutical R&D

#artificialintelligence

The level of #innovation in #pharmaceutical and #biotechnology R&D over the last few decades has been outstanding. We see that while disease management has radically changed and innovated, many experts in the industry as well as investors admit that the industry is slow to adopt to the new digital era and open up to the use of advanced technology to help with the #digitalization of #clinical research. The goal of a biotechnology company is to develop innovative medicines that demonstrate differentiated treatment opportunities while reducing the cost and time to market and maximizing return on investment. Recent Deloitte's analysis of return on pharmaceutical R&D investments for a cohort of 12 large biopharma companies shows a sustained decline from 10.1 percent in 2010 to 3.2 percent in 2017. Fortunately, new digital #technologies are coming out and aim to optimize the clinical development process, and more broadly the entire R&D value chain.