Accessing legal help in Bangladesh is hard. People face high fees, complex legal language, a shortage of lawyers, and millions of unresolved court cases. Generative AI models like OpenAI GPT-4.1 Mini, Gemini 2.0 Flash, Meta Llama 3 70B, and DeepSeek R1 could potentially democratize legal assistance by providing quick and affordable legal advice. In this study, we collected 250 authentic legal questions from the Facebook group "Know Your Rights," where verified legal experts regularly provide authoritative answers. These questions were subsequently submitted to four four advanced AI models and responses were generated using a consistent, standardized prompt. A comprehensive dual evaluation framework was employed, in which a state-of-the-art LLM model served as a judge, assessing each AI-generated response across four critical dimensions: factual accuracy, legal appropriateness, completeness, and clarity. Following this, the same set of questions was evaluated by three licensed Bangladeshi legal professionals according to the same criteria. In addition, automated evaluation metrics, including BLEU scores, were applied to assess response similarity. Our findings reveal a complex landscape where AI models frequently generate high-quality, well-structured legal responses but also produce dangerous misinformation, including fabricated case citations, incorrect legal procedures, and potentially harmful advice. These results underscore the critical need for rigorous expert validation and comprehensive safeguards before AI systems can be safely deployed for legal consultation in Bangladesh.

Inference-time steering aims to alter a large language model's (LLM's) responses without changing its parameters, but a central challenge is identifying the internal modules that most strongly govern the target behavior. Existing approaches often rely on simplistic cues or ad hoc heuristics, leading to suboptimal or unintended effects. We introduce REAL, a framework for identifying behavior-relevant modules (attention heads or layers) in Transformer models. For each module, REAL trains a vector-quantized autoencoder (VQ-AE) on its hidden activations and uses a shared, learnable codebook to partition the latent space into behavior-relevant and behavior-irrelevant subspaces. REAL quantifies a module's behavioral relevance by how well its VQ-AE encodings discriminate behavior-aligned from behavior-violating responses via a binary classification metric; this score guides both module selection and steering strength. We evaluate REAL across eight LLMs from the Llama and Qwen families and nine datasets spanning truthfulness enhancement, open-domain QA under knowledge conflicts, and general alignment tasks. REAL enables more effective inference-time interventions, achieving an average relative improvement of 20% (up to 81.5%) over the ITI method on truthfulness steering. In addition, the modules selected by REAL exhibit strong zero-shot generalization in cross-domain truthfulness-steering scenarios.

With the growing prevalence of large language models (LLMs), the safety of LLMs has raised significant concerns. However, there is still a lack of definitive standards for evaluating their safety due to the subjective nature of current safety benchmarks. To address this gap, we conducted the first exploration of LLMs' safety evaluation from a legal perspective by proposing the SafeLawBench benchmark. SafeLawBench categorizes safety risks into three levels based on legal standards, providing a systematic and comprehensive framework for evaluation. It comprises 24,860 multi-choice questions and 1,106 open-domain question-answering (QA) tasks. Our evaluation included 2 closed-source LLMs and 18 open-source LLMs using zero-shot and few-shot prompting, highlighting the safety features of each model. We also evaluated the LLMs' safety-related reasoning stability and refusal behavior. Additionally, we found that a majority voting mechanism can enhance model performance. Notably, even leading SOTA models like Claude-3.5-Sonnet and GPT-4o have not exceeded 80.5% accuracy in multi-choice tasks on SafeLawBench, while the average accuracy of 20 LLMs remains at 68.8\%. We urge the community to prioritize research on the safety of LLMs.

Several AI chatbots designed for fantasy and sexual role-playing conversations are leaking user prompts to the web in almost real time, new research seen by WIRED shows. Some of the leaked data shows people creating conversations detailing child sexual abuse, according to the research. Conversations with generative AI chatbots are near instantaneous--you type a prompt and the AI responds. If the systems are configured improperly, however, this can lead to chats being exposed. In March, researchers at the security firm UpGuard discovered around 400 exposed AI systems while scanning the web looking for misconfigurations.

The application of AI in education has gained widespread attention for its potential to enhance learning experiences across disciplines, including psychology [1, 2]. In the context of investigative interviewing, especially when questioning suspected child victims, AI offers a promising alternative to traditional training approaches. These conventional methods, often delivered through short workshops, fail to provide the hands-on practice, feedback, and continuous engagement needed for interviewers to master best practices in questioning child victims [3, 4]. Research has shown that while best practices recommend open-ended questions and discourage leading or suggestive queries [5, 6], many interviewers still struggle to implement these techniques effectively during real-world investigations [7]. The adoption of AI-powered child avatars provides a valuable solution, enabling Child Protective Services (CPS) workers to engage in realistic practice sessions without the ethical dilemmas associated with using real children, while simultaneously offering personalized feedback on their performance [8]. Our current system leverages advanced AI techniques within a structured virtual environment to train professionals in investigative interviewing. Specifically, this system integrates the Unity Engine to generate virtual avatars. Despite the potential advantages of our AI-based training system, its effectiveness largely depends on the perceived realism and fidelity of the virtual avatars used in these simulations [9]. Based on our findings, we observed that avatars generated using Generative Adversarial Networks (GANs) demonstrated higher levels of realism compared to those created with the Unity Engine in several key aspects [10].

For the past two years, millions of people searching for child abuse videos on Pornhub's UK website have been interrupted. Each of the 4.4 million times someone has typed in words or phrases linked to abuse, a warning message has blocked the page, saying that kind of content is illegal. And in half the cases, a chatbot has also pointed people to where they can seek help. The warning message and chatbot were deployed by Pornhub as part of a trial program, conducted with two UK-based child protection organizations, to find out whether people could be nudged away from looking for illegal material with small interventions. A new report analyzing the test, shared exclusively with WIRED, says the pop-ups led to a decrease in the number of searches for child sexual abuse material (CSAM) and saw scores of people seek support for their behavior.