Will the US Send Alleged Fuel Smugglers to Mexico? Mexico's extradition request for several American citizens accused of smuggling stolen oil exposes an imbalance in the US-Mexico bilateral security relationship.
AI agents that autonomously interact with external tools and environments have shown great promise across real-world applications. However, their reliance on external data exposes them to serious indirect prompt injection attacks, where malicious instructions embedded in third-party content hijack agent behaviors. To mitigate this threat, a growing number of defenses have been proposed and evaluated under existing agent security benchmarks. These benchmarks provide…
AI is moving from domain-specific autonomy in closed, predictable settings to large-language-model-driven agents that plan and act in open, cross-organizational environments. As a result, the cybersecurity risk landscape is changing in fundamental ways. Agentic AI systems can plan, act, collaborate, and persist over time, functioning as participants in complex socio-technical ecosystems rather than as isolated software components. Although recent work has strengthened…
Agentic AI systems built around large language models (LLMs) are moving away from closed, single-model frameworks and toward open ecosystems that connect a variety of agents, external tools, and resources. The Model Context Protocol (MCP) has emerged as a standard to unify tool access, allowing agents to discover, invoke, and coordinate with tools more flexibly. However, as MCP becomes more widely adopted, it also brings a new set of security and privacy challenges. These…
ABSTRACT The post–1945 international order is in flux. Trump's return to the White House in 2025 marks a critical juncture, reinforcing a structural transformation characterised by intensifying US–China rivalry, the weaponisation of economic statecraft and the fragmentation of multilateralism. Middle powers—positioned at the intersection of Washington's security imperatives and Beijing's economic gravitational pull—face mounting strategic challenges, but with the requisite…
General Tchiani accuses France, Benin and the Ivory Coast of links to attack near Niamey’s airport and thanks Russian troops for defence efforts Heavy security has been deployed around the main airport in Niger’s capital, Niamey, after overnight gunfire and explosions that the country’s military ruler blamed without evidence on France, Benin and Côte d’Ivoire. The shooting and detonations began shortly after midnight on Wednesday, according to residents of a neighbourhood…
As China emerges as a leading power in artificial intelligence (AI), its views on the ethical use of AI in warfare are increasingly consequential for global security. Yet these perspectives remain underexamined. This article offers the first systematic mapping of China’s domestic discourse on military AI ethics, drawing on official, military, and academic sources. It identifies three defining principles: the de-universalization of ethics in favour of sovereignty-based…
The rapid advancement and widespread adoption of generative artificial intelligence (GenAI) and large language models (LLMs) has been accompanied by the emergence of new security vulnerabilities and challenges, such as jailbreaking and other prompt injection attacks. These maliciously crafted inputs can exploit LLMs, causing data leaks, unauthorized actions, or compromised outputs, for instance. As both offensive and defensive prompt injection techniques evolve quickly, a…
This paper contributes to the scholarly debate on secondary states’ agency in the context of US–China technological competition by examining how members of the Five Eyes (FVEY) intelligence alliance, excluding the United States, responded to perceived security risks posed by Huawei’s 5G infrastructure. While all FVEY states ultimately excluded Huawei from national networks, the pathways to this outcome diverged markedly, reflecting differences in threat perception…
The Model Context Protocol (MCP) has emerged as a de facto standard for integrating Large Language Models with external tools, yet no formal security analysis of the protocol specification exists. We present the first rigorous security analysis of MCP's architectural design, identifying three fundamental protocol-level vulnerabilities: (1) absence of capability attestation allowing servers to claim arbitrary permissions, (2) bidirectional sampling without origin…
The proliferation of agentic AI coding assistants, including Claude Code, GitHub Copilot, Cursor, and emerging skill-based architectures, has fundamentally transformed software development workflows. These systems leverage Large Language Models (LLMs) integrated with external tools, file systems, and shell access through protocols like the Model Context Protocol (MCP). However, this expanded capability surface introduces critical security vulnerabilities. In this…
This article is written in response to Navigating the Divide: Why the U.S. Navy Must Resume FONOPs in Indonesia by Patrick Panjeti, Military Fellow at the CSIS Defence and Security Department. 1. Introduction: Misrepresenting Indonesia’s Legal Commitments In his essay “Navigating the Divide,” Patrick Panjeti presents a provocative and fictional scenario set in 2027: a
On the Radar: Mexico Responds to US Pressure Violent gang reprisals in Guatemala, a symbolic push for justice in Colombia, and mounting pressure on Mexico from the United States are in the spotlight this week. Dive into the latest developments shaping the region’s security landscape.
Large Language Model (LLM)-based question-answering systems offer significant potential for automating customer support and internal knowledge access in small businesses, yet their practical deployment remains challenging due to infrastructure costs, engineering complexity, and security risks, particularly in retrieval-augmented generation (RAG)-based settings. This paper presents an industry case study of an open-source, multi-tenant platform that enables small businesses…
Navigation agents powered by large language models (LLMs) convert natural language instructions into executable plans and actions. Compared to text-based applications, their security is far more critical: a successful prompt injection attack does not just alter outputs but can directly misguide physical navigation, leading to unsafe routes, mission failure, or real-world harm. Despite this high-stakes setting, the vulnerability of navigation agents to prompt injection…
Large foundation models are integrated into Computer Use Agents (CUAs), enabling autonomous interaction with operating systems through graphical user interfaces (GUIs) to perform complex tasks. This autonomy introduces serious security risks: malicious instructions or visual prompt injections can trigger unsafe reasoning and cause harmful system-level actions. Existing defenses, such as detection-based blocking, prevent damage but often abort tasks prematurely, reducing…
AI agents are autonomous systems that combine LLMs with external tools to solve complex tasks. While such tools extend capability, improper tool permissions introduce security risks such as indirect prompt injection and tool misuse. We characterize these failures as unbalanced tool-driven agency. Agents may retain unnecessary permissions (excessive agency) or fail to invoke required tools (insufficient agency), amplifying the attack surface and reducing performance. We…
In the contemporary political landscape, two major ideological approaches are commonly identified: liberalism and conservatism. Liberalism tends to be idealistic and equality-oriented, placing emphasis on freedom, human rights, and international cooperation and dialogue. Conservatism, by contrast, is more realistic and order-oriented, prioritizing national interest and security. Although it can sometimes become self-righteous, it generally maintains a pragmatic outlook and…
The rise of AI agent frameworks has introduced agent skills, modular packages containing instructions and executable code that dynamically extend agent capabilities. While this architecture enables powerful customization, skills execute with implicit trust and minimal vetting, creating a significant yet uncharacterized attack surface. We conduct the first large-scale empirical security analysis of this emerging ecosystem, collecting 42,447 skills from two major marketplaces…
MANILA – The foreign ministers of security allies the Philippines and Japan on Thursday agreed to further refine provisions of a military pact to allow smoother exchanges of troops and equipment as China’s harassment in the South China Sea has shown no signs of easing. Japan Foreign Minister Motegi Toshimitsu and his Filipino counterpart Ma
While LLM-based agents can interact with environments via invoking external tools, their expanded capabilities also amplify security risks. Monitoring step-level tool invocation behaviors in real time and proactively intervening before unsafe execution is critical for agent deployment, yet remains under-explored. In this work, we first construct TS-Bench, a novel benchmark for step-level tool invocation safety detection in LLM agents. We then develop a guardrail model…
AI agents are vulnerable to prompt injection attacks, where malicious content hijacks agent behavior. Among proposed defenses, architectural isolation provides the strongest guarantees by strictly separating trusted task planning from untrusted environment observations. However, applying this design to Computer Use Agents (CUAs), which automate tasks by viewing screens and executing actions, presents a fundamental challenge. Current agents require continuous observation of…
In 2023, Alfonso Camacho-Martinez wrote, “By Weakening the Military, Colombia’s Petro Imperils His Hopes for Peace,” where he argued that President Gustavo Petro has significantly weakened the military while in office. Three years later, in the midst of heightened tensions between the United States and some Latin American countries, we asked Alfonso to revisit his arguments. Image: U.S. National GuardIn your 2023 article, “By Weakening the Military, Colombia’s Petro Imperils…
On January 13, President and CEO of the Inter-American Dialogue Rebecca Bill Chavez spoke with DW News to discuss why China and Russia are closely watching Trump’s threats to Cuba.
It’s early days, but the Trump administration’s approach to a post-President Nicolás Maduro Venezuela is quickly coming into focus. This is not externally imposed democracy building but a regime management project that relies on compliant regime insiders, the existing state security apparatus, and offshore coercive leverage. As I argued previously in these pages, removing the apex of an authoritarian pyramid does not dissolve the pyramid. What changes is who gets to set the…
Large Language Models have emerged as transformative tools for Security Operations Centers, enabling automated log analysis, phishing triage, and malware explanation; however, deployment in adversarial cybersecurity environments exposes critical vulnerabilities to prompt injection attacks where malicious instructions embedded in security artifacts manipulate model behavior. This paper introduces SecureCAI, a novel defense framework extending Constitutional AI principles with…
Hint-based faithfulness evaluations have established that Large Reasoning Models (LRMs) may not say what they think: they do not always volunteer information about how key parts of the input (e.g. answer hints) influence their reasoning. Yet, these evaluations also fail to specify what models should do when confronted with hints or other unusual prompt content -- even though versions of such instructions are standard security measures (e.g. for countering prompt injections)…
Large language models (LLMs) are increasingly deployed in security-sensitive applications, where they must follow system- or developer-specified instructions that define the intended task behavior, while completing benign user requests. When adversarial instructions appear in user queries or externally retrieved content, models may override intended logic. Recent defenses rely on supervised fine-tuning with benign and malicious labels. Although these methods achieve high…
The study examines Pakistan's artificial intelligence (AI) strategy for national security and defence. It identifies AI trends and effects using qualitative data from 105 military, intelligence, cybersecurity, and government specialists. Thematic research indicates AI improves military decision-making, intelligence synthesis, surveillance, and real-time risk assessment. Participants emphasised the importance of AI in military resource allocation, target precision, predictive…