Results for “World”

OnCuba News Erian Pena

Returning with Frank Hinrichs to the splendor of Tropicana

He is one of the leading experts on the history of the cabaret and possesses the world’s largest collection of related objects. La entrada Returning with Frank Hinrichs to the splendor of Tropicana se publicó primero en OnCubaNews English.

Read at OnCuba News

arXiv (research papers) Zhibo Liang SI

SIEVE: Selective Integrity Verification and Escalation for Defending LLM Agents against Indirect Prompt Injection

Large Language Models (LLMs) are increasingly used as the core of agentic systems due to their strong reasoning, planning, and tool-use capabilities. By interacting with external environments, LLM agents can execute real-world tasks on behalf of users rather than merely generate text. This expanded capability also amplifies the threat of indirect prompt injection (IPI), where malicious external content can manipulate agent behavior and trigger unauthorized actions, privacy…

Read at arXiv (research papers)

Journal articles (OpenAlex) Alexander D. Barder · Global Studies Quarterly Geopolitics

“Things Just Don’t Make Sense”: Global Conspiracy Theories and Fascist Social Reality

Abstract This research note explores how global conspiracy theories contribute to the formation of fascistic representations of social reality. Instead of dismissing fascist worldviews as merely irrational or pathological, I focus on how their nonsensical form actively generates meaning. Drawing on Gilles Deleuze’s philosophy of sense developed in The Logic of Sense, I show that nonsense is not the absence of meaning but a creative and productive force that reshapes how the…

Read at Journal articles (OpenAlex)

arXiv (research papers) Omar Farooq Khan Suri SI

Securing Large Language Models (LLMs) from Prompt Injection Attacks

Large Language Models (LLMs) are increasingly being deployed in real-world applications, but their flexibility exposes them to prompt injection attacks. These attacks leverage the model's instruction-following ability to make it perform malicious tasks. Recent work has proposed JATMO, a task-specific fine-tuning approach that trains non-instruction-tuned base models to perform a single function, thereby reducing susceptibility to adversarial instructions. In this study, we…

Read at arXiv (research papers)

Journal articles (OpenAlex) Matěj Bílý · Europe Asia Studies Geopolitics

A Socialist CoCom? Warsaw Pact Export Controls in the Late 1980s

In 1985, the USSR proposed multilateral export controls for the Warsaw Pact in order to protect the ‘scientific-technological’ and ‘military-economic’ potential of its member states. In its aim to identify research, technologies and goods that members should be prohibited from transferring to partners in the West and the developing world, this proposal was analogous to the West's CoCom embargo. Using documents from two former Warsaw Pact countries, the German Democratic…

Read at Journal articles (OpenAlex)

Journal articles (OpenAlex) Guilherme Cavalcante Silva · Tapuya Latin American Science Technology and Society Geopolitics

Regional integration as a remedy against technological dependency? The contradictions in the calls for regional integration in Latin American AI strategies

The rise of regional integration and South-South cooperation as an imperative in international relations and trade has been widely discussed, especially with the strengthening of BRICS+ (Brazil, Russia, India, China, South Africa), and Pan-African movements. The idea of a world with multiple powers giving rise to diverse social and economic imperatives for technoscientific innovation is finding traction among scholars, including researchers in science and technology studies…

Read at Journal articles (OpenAlex)

arXiv (research papers) Kaiyuan Zhang SI

BrowseSafe: Understanding and Preventing Prompt Injection Within AI Browser Agents

The integration of artificial intelligence (AI) agents into web browsers introduces security challenges that go beyond traditional web application threat models. Prior work has identified prompt injection as a new attack vector for web agents, yet the resulting impact within real-world environments remains insufficiently understood. In this work, we examine the landscape of prompt injection attacks and synthesize a benchmark of attacks embedded in realistic HTML payloads…

Read at arXiv (research papers)

Simon Willison SI

Claude Opus 4.5, and why evaluating new LLMs is increasingly difficult

Anthropic released Claude Opus 4.5 this morning, which they call "best model in the world for coding, agents, and computer use". This is their attempt to retake the crown for best coding model after significant challenges from OpenAI's GPT-5.1-Codex-Max and Google's Gemini 3, both released within the past week! The core characteristics of Opus 4.5 are a 200,000 token context (same as Sonnet), 64,000 token output limit (also the same as Sonnet), and a March 2025 "reliable…

Read at Simon Willison

arXiv (research papers) Aram Vardanyan SI

Building Browser Agents: Architecture, Security, and Practical Solutions

Browser agents enable autonomous web interaction but face critical reliability and security challenges in production. This paper presents findings from building and operating a production browser agent. The analysis examines where current approaches fail and what prevents safe autonomous operation. The fundamental insight: model capability does not limit agent performance; architectural decisions determine success or failure. Security analysis of real-world incidents reveals…

Read at arXiv (research papers)