The Future of AI Research report, published in March 2025, aims to clearly identify the trajectory of AI research in a structured way. The report was led by outgoing AAAI President Francesca Rossi and covers 17 different AI topics . Members of the report team, and other selected AI practitioners, are taking part in a series of video panel discussions covering selected chapters from the report. In the fifth discussion in the collection, the three panellists tackle the topic of AI agents. How multi-agent systems evolved from rule-based systems to complex cooperative frameworks built on generative AI, and what is really different in the modern notion of an agentic AI system.

Autonomous agentic AI systems powered by vision-language models (VLMs) are rapidly advancing toward real-world deployment, yet their cross-modal reasoning capabilities introduce new attack surfaces for adversarial manipulation that exploit semantic reasoning across modalities. Existing adversarial attacks typically rely on visible pixel perturbations or require privileged model or environment access, making them impractical for stealthy, real-world exploitation. We introduce TRAP, a novel generative adversarial framework that manipulates the agent's decision-making using diffusion-based semantic injections into the vision-language embedding space. Our method combines negative prompt-based degradation with positive semantic optimization, guided by a Siamese semantic network and layout-aware spatial masking. Without requiring access to model internals, TRAP produces visually natural images yet induces consistent selection biases in agentic AI systems. We evaluate TRAP on the Microsoft Common Objects in Context (COCO) dataset, building multi-candidate decision scenarios. Across these scenarios, TRAP consistently induces decision-level preference redirection on leading models, including LLaVA-34B, Gemma3, GPT-4o, and Mistral-3.2,

LLMs excel at predictive tasks and complex reasoning tasks, but many high-value deployments rely on decisions under uncertainty, for example, which tool to call, which expert to consult, or how many resources to invest. While the usefulness and feasibility of Bayesian approaches remain unclear for LLM inference, this position paper argues that the control layer of an agentic AI system (that orchestrates LLMs and tools) is a clear case where Bayesian principles should shine. Bayesian decision theory provides a framework for agentic systems that can help to maintain beliefs over task-relevant latent quantities, to update these beliefs from observed agentic and human-AI interactions, and to choose actions. Making LLMs themselves explicitly Bayesian belief-updating engines remains computationally intensive and conceptually nontrivial as a general modeling target. In contrast, this paper argues that coherent decision-making requires Bayesian principles at the orchestration level of the agentic system, not necessarily the LLM agent parameters. This paper articulates practical properties for Bayesian control that fit modern agentic AI systems and human-AI collaboration, and provides concrete examples and design patterns to illustrate how calibrated beliefs and utility-aware policies can improve agentic AI orchestration.

The rapid deployment of large language model (LLM)-based agents introduces a new class of risks, driven by their capacity for autonomous planning, multi-step tool integration, and emergent interactions. It raises some risk factors for existing governance approaches as they remain fragmented: Existing frameworks are either static taxonomies driven; however, they lack an integrated end-to-end pipeline from risk identification to operational assurance, especially for an agentic platform. We propose AGENTSAFE, a practical governance framework for LLM-based agentic systems. The framework operationalises the AI Risk Repository into design, runtime, and audit controls, offering a governance framework for risk identification and assurance. The proposed framework, AGENTSAFE, profiles agentic loops (plan -> act -> observe -> reflect) and toolchains, and maps risks onto structured taxonomies extended with agent-specific vulnerabilities. It introduces safeguards that constrain risky behaviours, escalates high-impact actions to human oversight, and evaluates systems through pre-deployment scenario banks spanning security, privacy, fairness, and systemic safety. During deployment, AGENTSAFE ensures continuous governance through semantic telemetry, dynamic authorization, anomaly detection, and interruptibility mechanisms. Provenance and accountability are reinforced through cryptographic tracing and organizational controls, enabling measurable, auditable assurance across the lifecycle of agentic AI systems. The key contributions of this paper are: (1) a unified governance framework that translates risk taxonomies into actionable design, runtime, and audit controls; (2) an Agent Safety Evaluation methodology that provides measurable pre-deployment assurance; and (3) a set of runtime governance and accountability mechanisms that institutionalise trust in agentic AI ecosystems.

The AI conversation can be traced as far back as Alan Turing's milestone paper published in 1950, which considered the fundamental question "Can machines think?" [1]. In 1956, AI got its name and mission as a scientific field at the first AI conference held at Dartmouth College [2]. Following AI's foundational period in the 1950s ~ 1970s, AI has evolved from early rule-based systems (1970s ~ 1990s), through classical machine learning and deep learning with neural networks (1990s ~ 2020s), to today's generative and agentic AI systems (since 2010s). Correspondingly, as a vital requirement of these systems, the reliability concept and concerns are also evolving, particularly in the interpretation of "required function" (see Table 1 in Chapter 10), based on the definition in standards like ISO 8402 "The ability of an item to perform a required function, under given environmental and operational conditions and for a stated period of time ". While a conventional AI system is concerned with providing stable and accurate classifications, predictions, or optimizations, a reliable generative AI system focuses on producing outputs that are trustworthy, consistent, safe, and contextually appropriate [3]. Building on both, a reliable agentic AI system should additionally conduct functions of reasoning, goal alignment, planning, safe adaption and interaction in dynamic and collaborative multi-agent contexts. The expansion of reliability concepts has introduced new challenges and research opportunities, as exemplified in Figure 1. In the following sections, we shed lights on these challenges and opportunities in building reliable AI systems, particularly, agentic AI systems.

The implementation of agentic AI systems has the potential of providing more helpful AI systems in a variety of applications. These systems work autonomously towards a defined goal with reduced external control. Despite their potential, one of their flaws is the insufficient reliability which makes them especially unsuitable for high-risk domains such as healthcare or process industry. Unreliable systems pose a risk in terms of unexpected behavior during operation and mitigation techniques are needed. In this work, we derive the main reliability challenges of agentic AI systems during operation based on their characteristics. We draw the connection to traditional AI systems and formulate a fundamental reliability challenge during operation which is inherent to traditional and agentic AI systems. As our main contribution, we propose a two-layered reliability monitoring framework for agentic AI systems which consists of a out-of-distribution detection layer for novel inputs and AI transparency layer to reveal internal operations. This two-layered monitoring approach gives a human operator the decision support which is needed to decide whether an output is potential unreliable or not and intervene. This framework provides a foundation for developing mitigation techniques to reduce risk stemming from uncertain reliability during operation.

This chapter bridges technical analysis and organizational preparedness by tracing the path from layered failure modes to reliability awareness in generative and agentic AI systems. We first introduce an 11-layer failure stack, a structured framework for identifying vulnerabilities ranging from hardware and power foundations to adaptive learning and agentic reasoning. Building on this, the chapter demonstrates how failures rarely occur in isolation but propagate across layers, creating cascading effects with systemic consequences. To complement this diagnostic lens, we develop the concept of awareness mapping: a maturity-oriented framework that quantifies how well individuals and organizations recognize reliability risks across the AI stack. Awareness is treated not only as a diagnostic score but also as a strategic input for AI governance, guiding improvement and resilience planning. By linking layered failures to awareness levels and further integrating this into Dependability-Centred Asset Management (DCAM), the chapter positions awareness mapping as both a measurement tool and a roadmap for trustworthy and sustainable AI deployment across mission-critical domains.

While AI agents have long been discussed and studied in computer science, today's Agentic AI systems are something new. We consider other definitions of Agentic AI and propose a new realist definition. Agentic AI is a software delivery mechanism, comparable to software as a service (SaaS), which puts an application to work autonomously in a complex enterprise setting. Recent advances in large language models (LLMs) as foundation models have driven excitement in Agentic AI. We note, however, that Agentic AI systems are primarily applications, not foundations, and so their success depends on validation by end users and principal stakeholders. The tools and techniques needed by the principal users to validate their applications are quite different from the tools and techniques used to evaluate foundation models. Ironically, with good validation measures in place, in many cases the foundation models can be replaced with much simpler, faster, and more interpretable models that handle core logic. When it comes to Agentic AI, validity is what you need. LLMs are one option that might achieve it.

Artificial intelligence (AI) agents are emerging as transformative tools in drug discovery, with the ability to autonomously reason, act, and learn through complicated research workflows. Building on large language models (LLMs) coupled with perception, computation, action, and memory tools, these agentic AI systems could integrate diverse biomedical data, execute tasks, carry out experiments via robotic platforms, and iteratively refine hypotheses in closed loops. We provide a conceptual and technical overview of agentic AI architectures, ranging from ReAct and Reflection to Supervisor and Swarm systems, and illustrate their applications across key stages of drug discovery, including literature synthesis, toxicity prediction, automated protocol generation, small-molecule synthesis, drug repurposing, and end-to-end decision-making. To our knowledge, this represents the first comprehensive work to present real-world implementations and quantifiable impacts of agentic AI systems deployed in operational drug discovery settings. Early implementations demonstrate substantial gains in speed, reproducibility, and scalability, compressing workflows that once took months into hours while maintaining scientific traceability. We discuss the current challenges related to data heterogeneity, system reliability, privacy, and benchmarking, and outline future directions towards technology in support of science and translation.

Agentic AI systems, which leverage multiple autonomous agents and Large Language Models (LLMs), are increasingly used to address complex, multi-step tasks. The safety, security, and functionality of these systems are critical, especially in high-stakes applications. However, the current ecosystem of inter-agent communication is fragmented, with protocols such as the Model Context Protocol (MCP) for tool access and the Agent-to-Agent (A2A) protocol for coordination being analyzed in isolation. This fragmentation creates a semantic gap that prevents the rigorous analysis of system properties and introduces risks such as architectural misalignment and exploitable coordination issues. To address these challenges, we introduce a modeling framework for agentic AI systems composed of two foundational models. The first, the host agent model, formalizes the top-level entity that interacts with the user, decomposes tasks, and orchestrates their execution by leveraging external agents and tools. The second, the task lifecycle model, details the states and transitions of individual sub-tasks from creation to completion, providing a fine-grained view of task management and error handling. Together, these models provide a unified semantic framework for reasoning about the behavior of multi-AI agent systems. Grounded in this framework, we define 17 properties for the host agent and 14 for the task lifecycle, categorized into liveness, safety, completeness, and fairness. Expressed in temporal logic, these properties enable formal verification of system behavior, detection of coordination edge cases, and prevention of deadlocks and security vulnerabilities. Through this effort, we introduce the first rigorously grounded, domain-agnostic framework for the systematic analysis, design, and deployment of correct, reliable, and robust agentic AI systems.