Can we learn more from data than existed in the generating process itself? Can new and useful information be constructed from merely applying deterministic transformations to existing data? Can the learnable content in data be evaluated without considering a downstream task? On these questions, Shannon information and Kolmogorov complexity come up nearly empty-handed, in part because they assume observers with unlimited computational capacity and fail to target the useful information content. In this work, we identify and exemplify three seeming paradoxes in information theory: (1) information cannot be increased by deterministic transformations; (2) information is independent of the order of data; (3) likelihood modeling is merely distribution matching. To shed light on the tension between these results and modern practice, and to quantify the value of data, we introduce epiplexity, a formalization of information capturing what computationally bounded observers can learn from data. Epiplexity captures the structural content in data while excluding time-bounded entropy, the random unpredictable content exemplified by pseudorandom number generators and chaotic dynamical systems. With these concepts, we demonstrate how information can be created with computation, how it depends on the ordering of the data, and how likelihood modeling can produce more complex programs than present in the data generating process itself. We also present practical procedures to estimate epiplexity which we show capture differences across data sources, track with downstream performance, and highlight dataset interventions that improve out-of-distribution generalization. In contrast to principles of model selection, epiplexity provides a theoretical foundation for data selection, guiding how to select, generate, or transform data for learning systems.

The rapid rise of large language models (LLMs) has shifted artificial intelligence (AI) research toward agentic systems, motivating the use of weaker and more flexible notions of agency. However, this shift raises key questions about the extent to which LLM-based agents replicate human strategic reasoning, particularly in game-theoretic settings. In this context, we examine the role of agentic sophistication in shaping artificial reasoners' performance by evaluating three agent designs: a simple game-theoretic model, an unstructured LLM-as-agent model, and an LLM integrated into a traditional agentic framework. Using guessing games as a testbed, we benchmarked these agents against human participants across general reasoning patterns and individual role-based objectives. Furthermore, we introduced obfuscated game scenarios to assess agents' ability to generalise beyond training distributions. Our analysis, covering over 2000 reasoning samples across 25 agent configurations, shows that human-inspired cognitive structures can enhance LLM agents' alignment with human strategic behaviour. Still, the relationship between agentic design complexity and human-likeness is non-linear, highlighting a critical dependence on underlying LLM capabilities and suggesting limits to simple architectural augmentation.

This study examines the lexical and syntactic interventions of human and LLM proofreading aimed at improving overall intelligibility in identical second language writings, and evaluates the consistency of outcomes across three LLMs (ChatGPT-4o, Llama3.1-8b, Deepseek-r1-8b). Findings show that both human and LLM proofreading enhance bigram lexical features, which may contribute to better coherence and contextual connectedness between adjacent words. However, LLM proofreading exhibits a more generative approach, extensively reworking vocabulary and sentence structures, such as employing more diverse and sophisticated vocabulary and incorporating a greater number of adjective modifiers in noun phrases. The proofreading outcomes are highly consistent in major lexical and syntactic features across the three models.

Vocabulary use is a fundamental aspect of second language (L2) proficiency. To date, its assessment by automated systems has typically examined the context-independent, or part-of-speech (PoS) related use of words. This paper introduces a novel approach to enable fine-grained vocabulary evaluation exploiting the precise use of words within a sentence. The scheme combines large language models (LLMs) with the English Vocabulary Profile (EVP). The EVP is a standard lexical resource that enables in-context vocabulary use to be linked with proficiency level. We evaluate the ability of LLMs to assign proficiency levels to individual words as they appear in L2 learner writing, addressing key challenges such as polysemy, contextual variation, and multi-word expressions. We compare LLMs to a PoS-based baseline. LLMs appear to exploit additional semantic information that yields improved performance. We also explore correlations between word-level proficiency and essay-level proficiency. Finally, the approach is applied to examine the consistency of the EVP proficiency levels. Results show that LLMs are well-suited for the task of vocabulary assessment.

This paper presents a holistic approach to attacker preference modeling from system-level audit logs using inverse reinforcement learning (IRL). Adversary modeling is an important capability in cybersecurity that lets defenders characterize behaviors of potential attackers, which enables attribution to known cyber adversary groups. Existing approaches rely on documenting an ever-evolving set of attacker tools and techniques to track known threat actors. Although attacks evolve constantly, attacker behavioral preferences are intrinsic and less volatile. Our approach learns the behavioral preferences of cyber adversaries from forensics data on their tools and techniques. We model the attacker as an expert decision-making agent with unknown behavioral preferences situated in a computer host. We leverage attack provenance graphs of audit logs to derive a state-action trajectory of the attack. We test our approach on open datasets of audit logs containing real attack data. Our results demonstrate for the first time that low-level forensics data can automatically reveal an adversary's subjective preferences, which serves as an additional dimension to modeling and documenting cyber adversaries. Attackers' preferences tend to be invariant despite their different tools and indicate predispositions that are inherent to the attacker. As such, these inferred preferences can potentially serve as unique behavioral signatures of attackers and improve threat attribution.

As distributed artificial intelligence (AI) and multi-agent architectures grow increasingly complex, the need for adaptive, context-aware routing becomes paramount. This paper introduces an enhanced, adaptive routing algorithm tailored for AI multi-agent networks, integrating priority-based cost functions and dynamic learning mechanisms. Building on an extended Dijkstra-based framework, we incorporate multi-faceted parameters such as task complexity, user request priority, agent capabilities, bandwidth, latency, load, model sophistication, and reliability. We further propose dynamically adaptive weighting factors, tuned via reinforcement learning (RL), to continuously evolve routing policies based on observed network performance. Additionally, heuristic filtering and hierarchical routing structures improve scalability and responsiveness. Our approach yields context-sensitive, load-aware, and priority-focused routing decisions that not only reduce latency for critical tasks but also optimize overall resource utilization, ultimately enhancing the robustness, flexibility, and efficiency of multi-agent systems.

The British engineering company Arup has confirmed it was the victim of a deepfake fraud after an employee was duped into sending HK 200m ( 20m) to criminals by an artificial intelligence-generated video call. Hong Kong police said in February that a worker at a then-unnamed company had been tricked into transferring vast sums by people on a hoax call "posing as senior officers of the company". Arup said in a statement that it was the company involved, confirming that at the beginning of the year it had "notified the police about an incident of fraud in Hong Kong". It confirmed that fake voices and images were used. It added: "Our financial stability and business operations were not affected and none of our internal systems were compromised."

At a time when suspicion of manipulated media has reached a new pitch of concern, the Princess of Wales photo furore underlines the sensitivity around image doctoring. Catherine was the subject of an image editing row in 2011 when Grazia adapted a photo of her on her wedding day – but that was before breakthroughs in artificial intelligence put everyone on edge. There has been a deluge of AI-generated deepfakes in recent years, from a video of Volodymyr Zelenskiy telling his soldiers to surrender, to explicit images of Taylor Swift. Historical examples of image manipulation can be clunky – from Argentine footballers clutching handbags to Stalin's missing underlings – but there is now an alarming credibility to AI-generated content. Catherine's attempts to adjust a family photo, amid frenzied social media speculation about her wellbeing, have run straight into widespread concerns about trust in images, text and audio in a year when half the world is going to the polls.

Artificial intelligence will make it difficult to spot whether emails are genuine or sent by scammers and malicious actors, including messages that ask computer users to reset their passwords, the UK's cybersecurity agency has warned. The National Cyber Security Centre (NCSC) said people would struggle to identify phishing messages – where users are tricked into handing over passwords or personal details – due to the sophistication of AI tools. Generative AI, the term for technology that can produce convincing text, voice and images from simple hand-typed prompts, has become widely available to the public through chatbots such as ChatGPT and free-to-use versions known as open source models. The NCSC, part of the GCHQ spy agency, said in its latest assessment of AI's impact on the cyber threats facing the UK that AI would "almost certainly" increase the volume of cyber-attacks and heighten their impact over the next two years. It said generative AI and large language models – the technology that underpins chatbots – will complicate efforts to identify different types of attack such as spoof messages and social engineering, the term for manipulating people to hand over confidential material.

More than a decade before an AI-powered chatbot could do your homework, help you make dinner or pass the bar exam, there was @Horse_ebooks. The primitive predecessor to today's chatbot renaissance began as a Twitter account in 2010, tweeting automated excerpts from ebooks that, decontextualized, took on unexpected and strangely poetic meanings. Purportedly a spambot, the account surfaced quotes from ebooks that went viral for their absurdist fragments – phrases like "Hello saxophone," "COULD THIS BE THE", and "Today we are lucky to be talking". It amassed more than 200,000 followers at its peak and now, despite being inactive for a decade, the account still holds 131,000 followers. Its most memorable quip – "everything happens so much" – still resonates today.