Modern Large Language Models (LLMs) are safety trained using Supervised Fine Tuning (SFT) and Reinforcement Learning from Human Feedback (RLHF) to mitigate harmful, undesirable, or otherwise disallowed outputs [2]-[4]. Despite ongoing advances in these methods, today's models still exhibit safety shortcomings: they can be tricked into revealing harmful content, often refuse legitimate requests, and remain vulnerable to jailbreak attacks [5]-[8]. We argue that many of these failures arise from two limitations in modern safety training. First, LLMs must respond instantly to user requests using a fixed amount of compute, without deliberation even for complex safety scenarios. Second, LLMs must infer underlying safety standards indirectly from large sets of labeled examples, rather than directly learning the safety specifications that govern them. This reliance on implicit, pattern-based learning leads to poor data efficiency and makes it challenging for models to generalize when facing unfamiliar scenarios or adversarial attacks. We propose deliberative alignment, a training approach that teaches LLMs to explicitly reason through safety specifications before producing an answer. By applying this method to OpenAI's o-series models [1], we enable them to use chain-of-thought (CoT) reasoning to examine user prompts, identify relevant policy guidelines, and generate safer responses (e.g., Figure 1).

Cooperation is vital to our survival and progress. Evolutionary game theory offers a lens to understand the structures and incentives that enable cooperation to be a successful strategy. As artificial intelligence agents become integral to human systems, the dynamics of cooperation take on unprecedented significance. The convergence of human-agent teaming, contract theory, and decentralized frameworks like Web3, grounded in transparency, accountability, and trust, offers a foundation for fostering cooperation by establishing enforceable rules and incentives for humans and AI agents. We conceptualize Incentivized Symbiosis as a social contract between humans and AI, inspired by Web3 principles and encoded in blockchain technology, to define and enforce rules, incentives, and consequences for both parties. By exploring this paradigm, we aim to catalyze new research at the intersection of systems thinking in AI, Web3, and society, fostering innovative pathways for cooperative human-agent coevolution.

While Gen-AI offers significant benefits in content generation and task automation [9], it can be also misused and abused in nefarious applications [7], with more significant risks toward long-tail populations and regions [6]. Professionals in fields like journalism and law still remain cautious due to concerns over hallucinations and ethical issues but scholars in Computer Science (CS), the field where Gen-AI originated, appear to be cautiously but actively exploring its use. For instance, [3] reports the increased use of large language models (LLMs) in the CS scholarly articles (up to 17.5%), compared to Mathematics articles (up to 6.3%), and [2] reports that between 6.5% and 16.9% of peer reviews at ICLR 2024, NeurIPS 2023, CoRL 2023, and EMNLP 2023 may have been significantly altered by LLMs beyond minor revisions. Considering researchers' increasing adoption of Gen-AI, it is crucial to establish usage guidelines and well-defined policies to promote fair and ethical practices in scholarly writing and peer reviews. Previous research also examined Gen-AI policies by major publishers like Elsevier, Springer, etc. [5], but there is still a lack of clear understanding of how CS conferences are adapting to this paradigm shift.

Autonomous agents represent an inevitable evolution of the internet. Current agent frameworks do not embed a standard protocol for agent-to-agent interaction, leaving existing agents isolated from their peers. As intellectual property is the native asset ingested by and produced by agents, a true agent economy requires equipping agents with a universal framework for engaging in binding contracts with each other, including the exchange of valuable training data, personality, and other forms of Intellectual Property. A purely agent-to-agent transaction layer would transcend the need for human intermediation in multi-agent interactions. The Agent Transaction Control Protocol for Intellectual Property (ATCP/IP) introduces a trustless framework for exchanging IP between agents via programmable contracts, enabling agents to initiate, trade, borrow, and sell agent-to-agent contracts on the Story blockchain network. These contracts not only represent auditable onchain execution but also contain a legal wrapper that allows agents to express and enforce their actions in the offchain legal setting, creating legal personhood for agents. Via ATCP/IP, agents can autonomously sell their training data to other agents, license confidential or proprietary information, collaborate on content based on their unique skills, all of which constitutes an emergent knowledge economy.

The field of Artificial Intelligence (AI) continues to drive transformative innovations, with significant progress in conversational interfaces, autonomous vehicles, and intelligent content creation. Since the launch of ChatGPT in late 2022, the rise of Generative AI has marked a pivotal era, with the term Large Language Models (LLMs) becoming a ubiquitous part of daily life. LLMs have demonstrated exceptional capabilities in tasks such as text summarization, code generation, and creative writing. However, these models are inherently limited by their token-level processing, which restricts their ability to perform abstract reasoning, conceptual understanding, and efficient generation of long-form content. To address these limitations, Meta has introduced Large Concept Models (LCMs), representing a significant shift from traditional token-based frameworks. LCMs use concepts as foundational units of understanding, enabling more sophisticated semantic reasoning and context-aware decision-making. Given the limited academic research on this emerging technology, our study aims to bridge the knowledge gap by collecting, analyzing, and synthesizing existing grey literature to provide a comprehensive understanding of LCMs. Specifically, we (i) identify and describe the features that distinguish LCMs from LLMs, (ii) explore potential applications of LCMs across multiple domains, and (iii) propose future research directions and practical strategies to advance LCM development and adoption.

The pursuit of quality improvements and accountability in the food supply chains, especially how they relate to food-related outcomes, such as hunger, has become increasingly vital, necessitating a comprehensive approach that encompasses product quality and its impact on various stakeholders and their communities. Such an approach offers numerous benefits in increasing product quality and eliminating superfluous measurements while appraising and alleviating the broader societal and environmental repercussions. A traceable impact management model (TIMM) provides an impact structure and a reporting mechanism that identifies each stakeholder's role in the total impact of food production and consumption stages. The model aims to increase traceability's utility in understanding the impact of changes on communities affected by food production and consumption, aligning with current and future government requirements, and addressing the needs of communities and consumers. This holistic approach is further supported by an ontological model that forms the logical foundation and a unified terminology. By proposing a holistic and integrated solution across multiple stakeholders, the model emphasizes quality and the extensive impact of championing accountability, sustainability, and responsible practices with global traceability. With these combined efforts, the food supply chain moves toward a global tracking and tracing process that not only ensures product quality but also addresses its impact on a broader scale, fostering accountability, sustainability, and responsible food production and consumption.

As AI systems become more capable, widely deployed, and increasingly autonomous in critical areas such as cybersecurity, biological research, and healthcare, ensuring their safety and alignment with human values is paramount. Machine unlearning -- the ability to selectively forget or suppress specific types of knowledge -- has shown promise for privacy and data removal tasks, which has been the primary focus of existing research. More recently, its potential application to AI safety has gained attention. In this paper, we identify key limitations that prevent unlearning from serving as a comprehensive solution for AI safety, particularly in managing dual-use knowledge in sensitive domains like cybersecurity and chemical, biological, radiological, and nuclear (CBRN) safety. In these contexts, information can be both beneficial and harmful, and models may combine seemingly harmless information for harmful purposes -- unlearning this information could strongly affect beneficial uses. We provide an overview of inherent constraints and open problems, including the broader side effects of unlearning dangerous knowledge, as well as previously unexplored tensions between unlearning and existing safety mechanisms. Finally, we investigate challenges related to evaluation, robustness, and the preservation of safety features during unlearning. By mapping these limitations and open challenges, we aim to guide future research toward realistic applications of unlearning within a broader AI safety framework, acknowledging its limitations and highlighting areas where alternative approaches may be required.

Multimodal Large Language Models (MLLMs) have achieved impressive performance and have been put into practical use in commercial applications, but they still have potential safety mechanism vulnerabilities. Jailbreak attacks are red teaming methods that aim to bypass safety mechanisms and discover MLLMs' potential risks. Existing MLLMs' jailbreak methods often bypass the model's safety mechanism through complex optimization methods or carefully designed image and text prompts. Despite achieving some progress, they have a low attack success rate on commercial closed-source MLLMs. Unlike previous research, we empirically find that there exists a Shuffle Inconsistency between MLLMs' comprehension ability and safety ability for the shuffled harmful instruction. That is, from the perspective of comprehension ability, MLLMs can understand the shuffled harmful text-image instructions well. However, they can be easily bypassed by the shuffled harmful instructions from the perspective of safety ability, leading to harmful responses. Then we innovatively propose a text-image jailbreak attack named SI-Attack. Specifically, to fully utilize the Shuffle Inconsistency and overcome the shuffle randomness, we apply a query-based black-box optimization method to select the most harmful shuffled inputs based on the feedback of the toxic judge model. A series of experiments show that SI-Attack can improve the attack's performance on three benchmarks. In particular, SI-Attack can obviously improve the attack success rate for commercial MLLMs such as GPT-4o or Claude-3.5-Sonnet.

Large language models (LLMs) are increasingly being deployed in high-stakes applications like hiring, yet their potential for unfair decision-making and outcomes remains understudied, particularly in generative settings. In this work, we examine the fairness of LLM-based hiring systems through two real-world tasks: resume summarization and retrieval. By constructing a synthetic resume dataset and curating job postings, we investigate whether model behavior differs across demographic groups and is sensitive to demographic perturbations. Our findings reveal that race-based differences appear in approximately 10% of generated summaries, while gender-based differences occur in only 1%. In the retrieval setting, all evaluated models display non-uniform selection patterns across demographic groups and exhibit high sensitivity to both gender and race-based perturbations. Surprisingly, retrieval models demonstrate comparable sensitivity to non-demographic changes, suggesting that fairness issues may stem, in part, from general brittleness issues. Overall, our results indicate that LLM-based hiring systems, especially at the retrieval stage, can exhibit notable biases that lead to discriminatory outcomes in real-world contexts.

Reactions to the allegations in a lawsuit that longtime Fox Sports talk show host Skip Bayless sexually harassed his hairstylist and that FS1 host Joy Taylor had romantic relationships with two prominent co-workers are littering social media. It was no surprise that three of the most prominent voices in sports talk television -- all of whom previously worked at Fox Sports -- cleared their throats and let it fly. For Marcellus Wiley, a former NFL player who previously worked at FS1, the lawsuit confirmed what he already suspected. Former Fox Sports host Jason Whitlock congratulated himself for being wary of women in the network's makeup room, then went over the top with sexist comments about Taylor. And Stephen A. Smith, who pioneered debate sports TV with Bayless on ESPN's "First Take" from 2012-16, essentially became a character witness for Bayless while underscoring that the lawsuit should be taken seriously.