Results for “work”

arXiv (research papers) Yanting Wang SI

Agent Against Agent: An Agentic System for Automatic Prompt Injection Red Teaming

Prompt injection poses significant security risks to LLM agents. Efficient and effective red-teaming is therefore critical, both for evaluating these risks and for collecting training data to improve defenses. Existing state-of-the-art prompt injection red-teaming methods primarily rely on reinforcement learning (RL), producing attacker models that often generalize poorly to new target LLMs. In this work, we develop PIMiner, an agentic system for prompt injection…

Read at arXiv (research papers)

Prensa Latina (Cuban state media) Juan Carlos Dominguez Taño

Vietnamese Publisher releases *Selected Works* of Fidel Castro

Hanoi, Aug 5 (Prensa Latina) Vietnam’s "The Truth" National Political Publishing House has completed the printing of 3,000 sets of *Fidel Castro Ruz: Selected Works*, announced its director and editor-in-chief, Vu Trong Lam, during a ceremony in Hanoi. The post Vietnamese Publisher releases *Selected Works* of Fidel Castro first appeared on Prensa Latina.

Read at Prensa Latina (Cuban state media)

arXiv (research papers) Longtao Guo SI

LoginTrap: Uncovering Task-Agnostic Phishing-Style Indirect Prompt Injection Attacks against LLM-based Web Agents

LLM-based web agents automate user tasks by observing webpages and executing browser actions on behalf of users. As these agents operate on real web services, login becomes a sensitive authentication boundary because it involves credentials and sensitive information. Existing work shows that malicious webpage content can manipulate web agent actions, but it has not fully examined whether such content can induce login and cause end-to-end private data leakage. We study this…

Read at arXiv (research papers)

arXiv (research papers) Xuebin Li SI

Breadcrumbing Search Agents: Per-Turn Scheming Over Long-Horizon Trajectories

LLM-based search agents are widely used for information-seeking tasks, but their reliance on external tool returns introduces a critical security risk: web content retrieved during execution is untrusted, exposing agents to prompt injection and goal hijacking. Prior work on search-agent safety primarily focuses on static web-content injection, but modern agents issue follow-up queries and cross-check competing sources, so a single injected page is often diluted or rejected…

Read at arXiv (research papers)

Article 19 natashas

Africa: National counterterrorism mechanisms must have independent oversight

On 27 and 28 July 2026, ARTICLE 19 Eastern Africa, the Network of African National Human Rights Institutions (NANHRI), and the International Institute for Justice and the Rule of Law (IIJ) held the in-person awareness workshop Strengthening National Human Rights Institutions (NHRIs) Engagement on Counterterrorism Oversight in Africa in Nairobi, Kenya. The workshop aimed to support peer-to-peer

Read at Article 19

Journal articles (OpenAlex) Lena Molnar, Sarah McCook · Social Sciences Geopolitics

“Everyone’s Moving, and Learning Together”: Young People and Practitioners Working with Collective Care to Create Futures Without Gender-Based Violence

Gender-based violence may be considered a failure or absence of care across the interpersonal, symbolic, and structural levels. Feminist scholars have called for closer attention to an ethic of care as one vital pathway to shared solidarity for social justice and relations of non-violence. Works from Black, Indigenous, queer and disability writers and activists have similarly illustrated the transformative potential of collective care for building community and ending…

Read at Journal articles (OpenAlex)

arXiv (research papers) Peichun Hua SI

Behavioral Skill Reconstruction: Reconstructing Hidden Functionality from LLM Agent Skills

Closed source agent skills may encode proprietary instructions, scripts, constants, and data. Providers may offer their capabilities as services while keeping the underlying packages hidden. Prior work focuses on prompt injection attacks that directly disclose these artifacts, and existing defenses accordingly aim to prevent such leakage. However, preventing file disclosure does not prevent users from recovering the functionality those files implement. This raises a…

Read at arXiv (research papers)

Prensa Latina (Cuban state media) Ana Luisa Brown

Fidel Castro seminar in Vietnam focuses Historical legacy

Hanoi, Aug 3 (Prensa Latina) Deputy director of the National Academy of Politics Ho Chi Minh, Nguyen Manh Hung, said today that Fidel Castro was a strong internationalist, an excellent theorist and a practical activist of the communist and international workers movement. The post Fidel Castro seminar in Vietnam focuses Historical legacy first appeared on Prensa Latina.

Read at Prensa Latina (Cuban state media)

Journal articles (OpenAlex) Joel Stillerman · Social Problems Geopolitics

Crime and symbolic boundaries in unsettled times: an analysis of middle-class responses to fear in Chile

Abstract Students of crime in the United States examine drug economies in poor communities, poor residents’ adaptive behaviors, and punitive crime policies. Research on Latin America, the most violent region in the world, focuses on criminal governance, violence in poor communities, and the popularity of punitive policies. However, few scholars have examined middle class residents’ experiences of crime. This article builds on work on middle classes and utilizes interviews…

Read at Journal articles (OpenAlex)

Journal articles (OpenAlex) Lukas Bullock, Malin Espersson · NORA - Nordic Journal of Feminist and Gender Research Geopolitics

“Gender Equality is Not a Neutral Topic and There is a Tension in That”: Approaching Gender Mainstreaming in Neoliberal Sweden

This article examines how Swedish public sector gender administrators position themselves in relation to gender equality as a political project. Since the 1990s, Sweden’s adoption of gender mainstreaming alongside neoliberal reforms has led to increased bureaucratization of gender equality work, with administrative gender experts tasked with monitoring and implementing depoliticized gender mainstreaming schemes across local and regional contexts. While existing scholarship…

Read at Journal articles (OpenAlex)

Journal articles (OpenAlex) BİLGİN GÜNGÖR, Sercan Hamza Bağlama · Middle East Critique Geopolitics

Rethinking Early Socialist Literature in Turkey: Kahraman Kız and the Prehistory of the Turkish Socialist Novel

The emergence of socialist aesthetics in Turkish literature corresponds to the early Second Constitutional Era, a time of limited liberalization that enabled socialist ideologies to surface. Figures such as Rasim Haşmet—hailed by Asım Bezirci as the first socialist poet and journalist—alongside Abdülaziz Mecdî Tolun, Muallim Cûdî, Ahmed Rıfkı Baba, and Bedik, produced foundational works in various periodicals. Among these is Kahraman Kız (1910), a serialized novel published…

Read at Journal articles (OpenAlex)

Atlantic Council Enrique Millán-Mejía and Raúl González-Pietrogiovanna Geopolitics

Colombia’s next president wants to work with the US. Here’s how Washington should respond.

The Trump administration can support the next phase of the relationship by presenting a real alternative to Chinese financing for critical projects.

Read at Atlantic Council

The Guardian Haiti Associated Press

‘Each day feels like the last’: Haitians with TPS on edge as confusion clouds US work status

US employers and Haitian immigrants live in constant uncertainty as status of work permits remains in limbo For more than a month, Bobby Kuchinsky arrived at his Florida restaurant each morning worried about whether he still had enough staff or if this might be the last shift for some of his Haitian workers, who have spent more than a decade as cooks, busboys and oyster shuckers. On Tuesday, he suspended six of them after they lost authorization to work in the US.

Read at The Guardian Haiti

Berkeley AI Research blog SI

From CUDA to MLX: How K-Search Brings Decades of Kernel Expertise to Apple Silicon

Figure 1: CUDA-to-MLX optimization translation map. CUDA optimization knowledge can be translated into architecture-native MLX strategies rather than copied instruction-for-instruction. We face a new epoch in computing. Hardware is changing rapidly — not just faster GPUs, but a growing range of chips from different vendors, each with its own architecture and often tailored to specific AI workloads. Software is changing just as fast, and AI coding tools now generate in…

Read at Berkeley AI Research blog

Asia Times Benjamin Stuart Reed South China Sea

Watching American power tentatively arrive in the Philippines

OYSTER BAY, PALAWAN – When American forces withdrew from Iraq in 2011, I watched it happen from Kuwait, where I worked as a contracted security guard. The last convoy crossed the berm that December — roughly 100 vehicles — and with it went the war in which I had served. Two years later, at a

Read at Asia Times

Rappler Bea Cupin South China Sea

PH presents submission on extended continental shelf in West Philippine Sea

The National Mapping and Resource Information Authority (NAMRIA) Extended Continental Shelf Technical Working Group spent 15 years working on the submission

Read at Rappler

arXiv (research papers) Arseny Kravchenko SI

APPA: Recoverable Information-Flow Control for Real-World LLM Agents

LLM agents deployed in practical workflows routinely mix private context, untrusted tool and web outputs, and external side effects. While information-flow control (IFC) provides structural defenses against prompt injection, data exfiltration, and confused-deputy attacks, conventional IFC relies on monotone taint tracking that either over-blocks benign operations or permanently strands downstream execution once an agent ingests unvetted data. We present APPA (Agentic…

Read at arXiv (research papers)

arXiv (research papers) Max Landauer SI

Just Testing, Move Along: Evasion of LLM-based System Log Interpretation by Prompt Injection

Large Language Models (LLMs) are increasingly integrated into Security Operations Center (SOC) workflows, where they support analysts in tasks such as the interpretation of system logs. However, the ability of LLMs to directly process untrusted textual input also introduces new attack surfaces. In particular, attackers can inject contextual information or explicit instructions into log entries in order to influence how malicious activity is interpreted by the model. Despite…

Read at arXiv (research papers)

arXiv (research papers) Zhaoxi Zhang SI

Are You Still the Agent I Authorized? Earned Authority under a Fixed Ceiling for Evolving Agents

Long-lived AI agents increasingly evolve after deployment by retaining experience, acquiring skills and tools, revising workflows, delegating work, and moving across task phases. This improves adaptation but creates a distinct authorization problem. Tool-enabled agents can turn model errors and prompt injections into consequential external actions; when evolution occurs under a live grant, the subject exercising that authority or the context in which it acts may no longer…

Read at arXiv (research papers)

The Guardian Haiti Michael Sainato

‘We’re human beings’: Haitians in the US fear family separation and deportation to violence

More than 330,000 Haitian immigrants in limbo as Trump administration seeks to end temporary protected status More than 330,000 Haitian immigrants in the US are set to lose their ability to legally live and work in the country as the Trump administration seeks to terminate their temporary protected status. An estimated 13,000 are nursing assistants, who care for 65,000 patients every day. “I get up at 5am every morning and I take two buses to work to clean the rooms of the…

Read at The Guardian Haiti

Berkeley AI Research blog SI

Teaching LLMs to Update Beliefs for Efficient Long-Horizon Interaction

Overview of ABBEL compared to traditional recursive summarization. Beliefs replace the full interaction history as the agent’s working context, and belief grading improves performance by supervising the contents of each belief state.. As task horizons grow, LLM contexts can’t scale forever. Self-summarization enables concise, interpretable contexts, but at a significant performance cost, especially for human assistance domains where high quality data is scarce, e.g…

Read at Berkeley AI Research blog

Journal articles (OpenAlex) Faisal Al-Asaad · Journal of Labor and Society Geopolitics

Waste: A Marxist Theory of War in the Work of Ali Kadri

Abstract This paper draws on the work of political theorist Ali Kadri to address the theoretical ‘problem’ of war in Marxist thought. Addressing this problem has presented Marxism with formidable challenges, most notably the question of clarifying the nature of war, both as a distinctive social phenomenon as well as a political struggle, under capitalism. In this regard, Kadri’s work on development, sovereignty, and imperialism in the Arab World provides an invaluable…

Read at Journal articles (OpenAlex)

Journal articles (OpenAlex) Jesse L Grainger · Political Studies Review Geopolitics

Beyond co-occurrence: Homonativism, boundary-making, and the coupling of sexual liberalism and nativism

Across Europe, political actors increasingly claim to defend sexual minorities while pursuing exclusionary agendas on immigration, integration, and national belonging. Recent work has identified this alignment as homonativism: the strategic incorporation of sexual-liberal stances towards homosexuality into nativist boundary-making, applied primarily to party rhetoric and electoral competition. This article extends homonativism to the mass level, asking how sexual-liberal and…

Read at Journal articles (OpenAlex)

The Guardian Myanmar Natasha May in Bangkok Geopolitics

‘We feel like we are failing’: Myanmar doctor speaks out as health system teeters on ‘brink of collapse’

Elle’s story is one of personal sacrifice, deciding to help people despite the risk of attack by the military. Her story reflects that of a healthcare system in peril When Myanmar’s junta seized power in a 2021 coup, Elle* and a fellow doctor fled to the jungle to provide aid to the resistance movement, one of 45,000 healthcare workers to lend their support. Elle does not regret her decision, she supports the pro-democracy movement and later married the doctor with whom she…

Read at The Guardian Myanmar

Journal articles (OpenAlex) Fathima Zehba M. P., Lakshmi Priya Rajendran, Mohammed Firoz C · Third World Quarterly Geopolitics

Gender and mobility: household dynamics, caregiving and patriarchal constraints shaping women’s movement in Kochi, India

Women’s mobility is shaped by gendered social norms, caregiving responsibilities and patriarchal expectations. This study examines how middle-aged women in urban India navigate mobility constraints produced by household gender dynamics and care work. Drawing on Bourdieu’s concept of habitus and Hochschild’s ‘second shift’, the analysis explores how women’s mobility choices are conditioned by internalised norms and the everyday demands of caregiving. Based on 25 in-depth…

Read at Journal articles (OpenAlex)

The Guardian Communism Editorial

The Guardian view on Chinese workers picking up the pen: poetry in motion | Editorial

A food delivery rider has won the country’s highest literary award. It is part of a growing trend for writing by labourers The hundreds of millions who have built modern China from the ground up – sweeping its streets, labouring in its factories and ferrying its goods – are often overlooked. But they refuse to go unheard. Last week, a food delivery rider won China’s top literary award. Wang Jibing, 56, first caught attention when a poem about his experiences went viral in…

Read at The Guardian Communism

arXiv (research papers) Gjergji Kasneci SI

The safety failures we are not instrumenting: a perspective on hidden safety-critical challenges in modern AI systems

Current AI safety discourse still focuses disproportionately on visible failures, including obvious harms, dramatic misuse, and hypothetical catastrophic scenarios. That focus is incomplete. In deployed systems, many of the most consequential failures are quieter: plausible rather than spectacular, distributed across components rather than localized in a single output, and normalized by workflows before they are recognized as hazards. We argue that a central safety challenge…

Read at arXiv (research papers)

arXiv (research papers) Akansha Shukla SI

Data Leakage Prevention in Agentic Applications via Preemptive Hardening

Agentic systems integrate LLM driven planning with interfaces to external tools, making data leakage and tool misuse feasible via instruction/data boundary failures and prompt injection attacks. Enforcing required controls consistently is particularly challenging in workflows spanning many codebases and heterogeneous agents. To address this challenge in multi agentic systems, we present a pre-deployment pipeline for scanning, hardening, and validation of agentic…

Read at arXiv (research papers)

arXiv (research papers) Jie Li SI

Trusted Credentials, Untrusted Behavior: Benchmarking LLM-Agent Security in High-Performance Computing

Large language model (LLM) agents are starting to take on routine work in high-performance computing (HPC), including monitoring Slurm jobs, diagnosing failed builds, inspecting simulation output, and coordinating scientific workflows. To do this work, an agent commonly acts under its user's credentials and inherits the user's access to files and the scheduler. This arrangement creates a failure mode that ordinary account-level controls do not capture. Adversarial…

Read at arXiv (research papers)

Article 19 natashas

ARTICLE 19 Eastern Africa: Annual Report 2025

Across Eastern Africa, democratic participation and civic freedoms continue to face profound challenges. Restrictions on media freedom, expanding surveillance infrastructures, and increasing regulation of online speech are reshaping the environment in which citizens exercise their fundamental rights. Within this evolving landscape, the work of ARTICLE 19 Eastern Africa remains more essential than ever. Throughout 2025,

Read at Article 19

Older stories