Vision-and-Language models such as CLIP have demonstrated remarkable effectiveness across a wide range of tasks. However, these models are typically trained on web-scale data, which can introduce inappropriate content and lead to the development of unsafe and biased behavior. This, in turn, hampers their applicability in sensitive and trustworthy contexts and could raise significant concern in their adoption. To overcome these limitations, we introduce a methodology to make Vision-and-Language models safer by removing their sensitivity to not-safe-for-work concepts. We show how this can be done by distilling from a large language model which converts between safe and unsafe sentences and which is fine-tuned starting from just 100 manually-curated pairs. We conduct extensive experiments on the resulting embedding space for both retrieval and text-to-image generation, where we show that our model can also be properly employed with pre-trained image generators. Our source code and trained models are available at: https://github.com/aimagelab/safe-clip.

Large language models (LLMs) can potentially democratize access to medical knowledge. While many efforts have been made to harness and improve LLMs' medical knowledge and reasoning capacities, the resulting models are either closed-source (e.g., PaLM, GPT-4) or limited in scale (<= 13B parameters), which restricts their abilities. In this work, we improve access to large-scale medical LLMs by releasing MEDITRON: a suite of open-source LLMs with 7B and 70B parameters adapted to the medical domain. MEDITRON builds on Llama-2 (through our adaptation of Nvidia's Megatron-LM distributed trainer), and extends pretraining on a comprehensively curated medical corpus, including selected PubMed articles, abstracts, and internationally-recognized medical guidelines. Evaluations using four major medical benchmarks show significant performance gains over several state-of-the-art baselines before and after task-specific finetuning. Overall, MEDITRON achieves a 6% absolute performance gain over the best public baseline in its parameter class and 3% over the strongest baseline we finetuned from Llama-2. Compared to closed-source LLMs, MEDITRON-70B outperforms GPT-3.5 and Med-PaLM and is within 5% of GPT-4 and 10% of Med-PaLM-2. We release our code for curating the medical pretraining corpus and the MEDITRON model weights to drive open-source development of more capable medical LLMs.

The rapidity with which generative AI has been adopted and advanced has raised legal and ethical questions related to the impact on artists rights, content production, data collection, privacy, accuracy of information, and intellectual property rights. Recent administrative and case law challenges have shown that generative AI software systems do not have independent intellectual property rights in the content that they generate. It remains to be seen whether human content creators can retain their intellectual property rights against generative AI software, its developers, operators, and owners for the misappropriation of the work of human creatives, given the metes and bounds of existing law. Early signs from various courts are mixed as to whether and to what degree the results generated by AI models meet the legal standards of infringement under existing law.

Machine learning models are deployed as a central component in decision making and policy operations with direct impact on individuals' lives. In order to act ethically and comply with government regulations, these models need to make fair decisions and protect the users' privacy. However, such requirements can come with decrease in models' performance compared to their potentially biased, privacy-leaking counterparts. Thus the trade-off between fairness, privacy and performance of ML models emerges, and practitioners need a way of quantifying this trade-off to enable deployment decisions. In this work we interpret this trade-off as a multi-objective optimization problem, and propose PFairDP, a pipeline that uses Bayesian optimization for discovery of Pareto-optimal points between fairness, privacy and utility of ML models. We show how PFairDP can be used to replicate known results that were achieved through manual constraint setting process. We further demonstrate effectiveness of PFairDP with experiments on multiple models and datasets.

Blockchain-empowered federated learning (FL) has provoked extensive research recently. Various blockchain-based federated learning algorithm, architecture and mechanism have been designed to solve issues like single point failure and data falsification brought by centralized FL paradigm. Moreover, it is easier to allocate incentives to nodes with the help of the blockchain. Various centralized federated learning frameworks like FedML, have emerged in the community to help boost the research on FL. However, decentralized blockchain-based federated learning framework is still missing, which cause inconvenience for researcher to reproduce or verify the algorithm performance based on blockchain. Inspired by the above issues, we have designed and developed a blockchain-based federated learning framework by embedding Ethereum network. This report will present the overall structure of this framework, which proposes a code practice paradigm for the combination of FL with blockchain and, at the same time, compatible with normal FL training task. In addition to implement some blockchain federated learning algorithms on smart contract to help execute a FL training, we also propose a model ownership authentication architecture based on blockchain and model watermarking to protect the intellectual property rights of models. These mechanism on blockchain shows an underlying support of blockchain for federated learning to provide a verifiable training, aggregation and incentive distribution procedure and thus we named this framework VeryFL (A Verify Federated Learninig Framework Embedded with Blockchain). The source code is avaliable on https://github.com/GTMLLab/VeryFL.

Federated Unlearning (FU) aims to delete specific training data from an ML model trained using Federated Learning (FL). We introduce QuickDrop, an efficient and original FU method that utilizes dataset distillation (DD) to accelerate unlearning and drastically reduces computational overhead compared to existing approaches. In QuickDrop, each client uses DD to generate a compact dataset representative of the original training dataset, called a distilled dataset, and uses this compact dataset during unlearning. To unlearn specific knowledge from the global model, QuickDrop has clients execute Stochastic Gradient Ascent with samples from the distilled datasets, thus significantly reducing computational overhead compared to conventional FU methods. We further increase the efficiency of QuickDrop by ingeniously integrating DD into the FL training process. By reusing the gradient updates produced during FL training for DD, the overhead of creating distilled datasets becomes close to negligible. Evaluations on three standard datasets show that, with comparable accuracy guarantees, QuickDrop reduces the duration of unlearning by 463.8x compared to model retraining from scratch and 65.1x compared to existing FU approaches. We also demonstrate the scalability of QuickDrop with 100 clients and show its effectiveness while handling multiple unlearning operations.

The study deployed four AI-detection tools to uncover possible evidence of hastily crafted, insincere apologies penned by an AI chatbot: ChatGPTZero, Undetectable.ai, The team collected apologies from posts and videos and transcribed the content, which was scanned through the four systems and then analyzed for the likelihood of AI generation. Overall percentages were then calculated, finding averages of each percentage found on different AI detector tools for that celebrity. Viral TikTok sensation Tiffany Gomas, whose unhinged reaction to another airline passenger spanned the internet, scored a combined 72 percent likelihood that AI-generated her video-taped Instagram apology. The transcribed text of the content was found to be 99 percent likely AI on Sapling and 45 percent likely AI on ChatGPTZero.

Big Data empowers the farming community with the information needed to optimize resource usage, increase productivity, and enhance the sustainability of agricultural practices. The use of Big Data in farming requires the collection and analysis of data from various sources such as sensors, satellites, and farmer surveys. While Big Data can provide the farming community with valuable insights and improve efficiency, there is significant concern regarding the security of this data as well as the privacy of the participants. Privacy regulations, such as the EU GDPR, the EU Code of Conduct on agricultural data sharing by contractual agreement, and the proposed EU AI law, have been created to address the issue of data privacy and provide specific guidelines on when and how data can be shared between organizations. To make confidential agricultural data widely available for Big Data analysis without violating the privacy of the data subjects, we consider privacy-preserving methods of data sharing in agriculture. Deep learning-based synthetic data generation has been proposed for privacy-preserving data sharing. However, there is a lack of compliance with documented data privacy policies in such privacy-preserving efforts. In this study, we propose a novel framework for enforcing privacy policy rules in privacy-preserving data generation algorithms. We explore several available agricultural codes of conduct, extract knowledge related to the privacy constraints in data, and use the extracted knowledge to define privacy bounds in a privacy-preserving generative model. We use our framework to generate synthetic agricultural data and present experimental results that demonstrate the utility of the synthetic dataset in downstream tasks. We also show that our framework can evade potential threats and secure data based on applicable regulatory policy rules.

While carbon accounting plays a fundamental role in our fight against climate change, it is not without its challenges. We begin the paper with a critique of the conventional carbon accounting practices, after which we proceed to introduce the E-liability carbon accounting methodology and Emissions Liability Management (ELM) originally proposed by Kaplan and Ramanna, highlighting their strengths. Recognizing the immense value of this novel approach for real-world carbon accounting improvement, we introduce a novel data-driven integrative framework that leverages AI and computation - the E-Liability Knowledge Graph framework - to achieve real-world implementation of the E-liability carbon accounting methodology. In addition to providing a path-to-implementation, our proposed framework brings clarity to the complex environmental interactions within supply chains, thus enabling better informed and more responsible decision-making. We analyze the implementation aspects of this framework and conclude with a discourse on the role of this AI-aided knowledge graph in ensuring the transparency and decarbonization of global supply chains.

As artificial intelligence (AI) is integrated into various services and systems in society, many companies and organizations have proposed AI principles, policies, and made the related commitments. Conversely, some have proposed the need for independent audits, arguing that the voluntary principles adopted by the developers and providers of AI services and systems insufficiently address risk. This policy recommendation summarizes the issues related to the auditing of AI services and systems and presents three recommendations for promoting AI auditing that contribute to sound AI governance. Recommendation1.Development of institutional design for AI audits. Recommendation2.Training human resources for AI audits. Recommendation3. Updating AI audits in accordance with technological progress. In this policy recommendation, AI is assumed to be that which recognizes and predicts data with the last chapter outlining how generative AI should be audited.