As the artificial intelligence (AI) supply chain grows more complex, AI systems and models are increasingly likely to incorporate externally-sourced ingredients such as datasets and other models. In such cases, determining whether or not an AI system or model complies with the EU AI Act will require gathering compliance-related metadata about both the AI system or model at-large as well as those externally-supplied ingredients. There must then be an analysis that looks across all of this metadata to render a prediction about the compliance of the overall AI system or model. Up until now, this process has not been automated. Thus, it has not been possible to make real-time compliance determinations in scenarios where doing so would be advantageous, such as the iterative workflows of today's AI developers, search and acquisition of AI ingredients on communities like Hugging Face, federated and continuous learning, and more. To address this shortcoming, we introduce a highly automated system for AI Act compliance analysis. This system has two key elements. First is an interlocking set of computational artifacts that capture compliance-related metadata about both: (1) the AI system or model at-large; (2) any constituent ingredients such as datasets and models. Second is an automated analysis algorithm that operates across those computational artifacts to render a run-time prediction about whether or not the overall AI system or model complies with the AI Act. Working together, these elements promise to enhance and accelerate AI Act compliance assessments.

Recently there has been a surge of interest in extending the success of large language models (LLMs) to graph modality, such as social networks and molecules. As LLMs are predominantly trained with 1D text data, most existing approaches adopt a graph neural network to represent a graph as a series of node tokens and feed these tokens to LLMs for graph-language alignment. Despite achieving some successes, existing approaches have overlooked the hierarchical structures that are inherent in graph data. Especially, in molecular graphs, the high-order structural information contains rich semantics of molecular functional groups, which encode crucial biochemical functionalities of the molecules. We establish a simple benchmark showing that neglecting the hierarchical information in graph tokenization will lead to subpar graph-language alignment and severe hallucination in generated outputs. To address this problem, we propose a novel strategy called HIerarchical GrapH Tokenization (HIGHT). HIGHT employs a hierarchical graph tokenizer that extracts and encodes the hierarchy of node, motif, and graph levels of informative tokens to improve the graph perception of LLMs. HIGHT also adopts an augmented graph-language supervised fine-tuning dataset, enriched with the hierarchical graph information, to further enhance the graph-language alignment. Extensive experiments on 7 molecule-centric benchmarks confirm the effectiveness of HIGHT in reducing hallucination by 40%, as well as significant improvements in various molecule-language downstream tasks.

Research on dialogue constructiveness assessment focuses on (i) analysing conversational factors that influence individuals to take specific actions, win debates, change their perspectives or broaden their open-mindedness and (ii) predicting constructive outcomes following dialogues for such use cases. These objectives can be achieved by training either interpretable feature-based models (which often involve costly human annotations) or neural models such as pre-trained language models (which have empirically shown higher task accuracy but lack interpretability). We propose a novel LLM feature-based framework that combines the strengths of feature-based and neural approaches while mitigating their downsides, in assessing dialogue constructiveness. The framework first defines a set of dataset-independent and interpretable linguistic features, which can be extracted by both prompting an LLM and simple heuristics. Such features are then used to train LLM feature-based models. We apply this framework to three datasets of dialogue constructiveness and find that our LLM feature-based models significantly outperform standard feature-based models and neural models, and tend to learn more robust prediction rules instead of relying on superficial shortcuts (as seen with neural models). Further, we demonstrate that interpreting these LLM feature-based models can yield valuable insights into what makes a dialogue constructive.

Recent studies show that image and video generation models can be prompted to reproduce copyrighted content from their training data, raising serious legal concerns around copyright infringement. Copyrighted characters, in particular, pose a difficult challenge for image generation services, with at least one lawsuit already awarding damages based on the generation of these characters. Yet, little research has empirically examined this issue. We conduct a systematic evaluation to fill this gap. First, we build CopyCat, an evaluation suite consisting of diverse copyrighted characters and a novel evaluation pipeline. Our evaluation considers both the detection of similarity to copyrighted characters and generated image's consistency with user input. Our evaluation systematically shows that both image and video generation models can still generate characters even if characters' names are not explicitly mentioned in the prompt, sometimes with only two generic keywords (e.g., prompting with "videogame, plumber" consistently generates Nintendo's Mario character). We then introduce techniques to semi-automatically identify such keywords or descriptions that trigger character generation. Using our evaluation suite, we study runtime mitigation strategies, including both existing methods and new strategies we propose. Our findings reveal that commonly employed strategies, such as prompt rewriting in the DALL-E system, are not sufficient as standalone guardrails. These strategies must be coupled with other approaches, like negative prompting, to effectively reduce the unintended generation of copyrighted characters. Our work provides empirical grounding to the discussion of copyright mitigation strategies and offers actionable insights for model deployers actively implementing them.

Large language models (LLMs) are increasingly being used in human-centered social scientific tasks, such as data annotation, synthetic data creation, and engaging in dialog. However, these tasks are highly subjective and dependent on human factors, such as one's environment, attitudes, beliefs, and lived experiences. Thus, employing LLMs (which do not have such human factors) in these tasks may result in a lack of variation in data, failing to reflect the diversity of human experiences. In this paper, we examine the role of prompting LLMs with human-like personas and asking the models to answer as if they were a specific human. This is done explicitly, with exact demographics, political beliefs, and lived experiences, or implicitly via names prevalent in specific populations. The LLM personas are then evaluated via (1) subjective annotation task (e.g., detecting toxicity) and (2) a belief generation task, where both tasks are known to vary across human factors. We examine the impact of explicit vs. implicit personas and investigate which human factors LLMs recognize and respond to. Results show that LLM personas show mixed results when reproducing known human biases, but generate generally fail to demonstrate implicit biases. We conclude that LLMs lack the intrinsic cognitive mechanisms of human thought, while capturing the statistical patterns of how people speak, which may restrict their effectiveness in complex social science applications.

Retrieval-Augmented Generation (RAG) improves pre-trained models by incorporating external knowledge at test time to enable customized adaptation. We study the risk of datastore leakage in Retrieval-In-Context RAG Language Models (LMs). We show that an adversary can exploit LMs' instruction-following capabilities to easily extract text data verbatim from the datastore of RAG systems built with instruction-tuned LMs via prompt injection. The vulnerability exists for a wide range of modern LMs that span Llama2, Mistral/Mixtral, Vicuna, SOLAR, WizardLM, Qwen1.5, and Platypus2, and the exploitability exacerbates as the model size scales up. Extending our study to production RAG models GPTs, we design an attack that can cause datastore leakage with a 100% success rate on 25 randomly selected customized GPTs with at most 2 queries, and we extract text data verbatim at a rate of 41% from a book of 77,000 words and 3% from a corpus of 1,569,000 words by prompting the GPTs with only 100 queries generated by themselves.

CryptoGPT: a 7B model competing with GPT-4 in a specific task -- The Impact of Automatic Annotation and Strategic Fine-Tuning via QLoRAIn this article, we present a method aimed at refining a dedicated LLM of reasonable quality with limited resources in an industrial setting via CryptoGPT. It is an LLM designed for financial news analysis for the cryptocurrency market in real-time. This project was launched in an industrial context. This model allows not only for the classification of financial information but also for providing comprehensive analysis. We refined different LLMs of the same size such as Mistral-7B and LLama-7B using semi-automatic annotation and compared them with various LLMs such as GPT-3.5 and GPT-4. Our goal is to find a balance among several needs: 1. Protecting data (by avoiding their transfer to external servers), 2. Limiting annotation cost and time, 3. Controlling the model's size (to manage deployment costs), and 4. Maintaining better analysis quality.

Formal verification (FV) has witnessed growing significance with current emerging program synthesis by the evolving large language models (LLMs). However, current formal verification mainly resorts to symbolic verifiers or hand-craft rules, resulting in limitations for extensive and flexible verification. On the other hand, formal languages for automated theorem proving, such as Isabelle, as another line of rigorous verification, are maintained with comprehensive rules and theorems. In this paper, we propose FVEL, an interactive Formal Verification Environment with LLMs. Specifically, FVEL transforms a given code to be verified into Isabelle, and then conducts verification via neural automated theorem proving with an LLM. The joined paradigm leverages the rigorous yet abundant formulated and organized rules in Isabelle and is also convenient for introducing and adjusting cutting-edge LLMs. To achieve this goal, we extract a large-scale FVELER3. The FVELER dataset includes code dependencies and verification processes that are formulated in Isabelle, containing 758 theories, 29,125 lemmas, and 200,646 proof steps in total with in-depth dependencies. We benchmark FVELER in the FVEL environment by first fine-tuning LLMs with FVELER and then evaluating them on Code2Inv and SV-COMP. The results show that FVEL with FVELER fine-tuned Llama3- 8B solves 17.39% (69 -> 81) more problems, and Mistral-7B 12% (75 -> 84) more problems in SV-COMP. And the proportion of proof errors is reduced. Project page: https://fveler.github.io/.

When users first found out about Adobe's new terms of service (which were quietly updated in February), there was an uproar. Adobe told users it could access their content "through both automated and manual methods" and use "techniques such as machine learning in order to improve [Adobe's] Services and Software." Many understood the update as the company forcing users to grant unlimited access to their work, for purposes of training Adobe's generative AI: Firefly. Late on Tuesday, Adobe issued a clarification: In an updated version of its terms of service agreement, it pledged not to train AI on its user content stored locally or in the cloud and gave users the option to opt-out of content analytics. Caught in the crossfire of intellectual property lawsuits, the ambiguous language used to previously update the terms shed light on a climate of acute skepticism among artists, many of whom over rely on Adobe for their work.

An enduring problem with today's generative artificial intelligence (AI) tools, like ChatGPT, is that they often confidently assert false information. Computer scientists call this behavior "hallucination," and it's a key barrier to AI's usefulness. Hallucinations have led to some embarrassing public slip-ups. In February, AirCanada was forced by a tribunal to honor a discount that its customer-support chatbot had mistakenly offered to a passenger. In May, Google was forced to make changes to its new "AI overviews" search feature, after the bot told some users that it was safe to eat rocks. And last June, two lawyers were fined 5,000 by a U.S. judge after one of them admitted he had used ChatGPT to help write a court filing.