Safety & alignment

We introduce NAILS (Normative Alignment of Recommender Systems via Internal Label Shift), a simple and scalable method for aligning recommendation outputs with target distributions over item-level attributes, such as categories. Recommender systems optimized solely for user engagement often fail to satisfy broader normative objectives, including fairness, diversity, and editorial values. NAILS modifies the user-conditional item distribution to induce a specified marginal distribution over attrib

Contemplative traditions have long guided ethical behavior and prosocial interaction, and recent work suggests that contemplative principles (e.g., mindfulness, compassion, non-dual reasoning) may offer a promising paradigm for aligning large language models (LLMs), improving cooperation and reducing ethical violations in LLM outputs. However, as new models, evaluation metrics, and benchmarks emerge rapidly, it remains challenging to systematically assess whether and how contemplative principles

Efficient waste segregation is critical for sustainable urban management and environmental governance. Existing automated systems are limited by single-modality visual processing, insufficient contextual understanding, and weak regulatory alignment. To address these issues, we propose a language-guided vision-AI framework that integrates vision-language models and multimodal large language models for joint visual-linguistic reasoning. This framework implements a visual question answering paradig

Network alignment identifies node correspondences across different networks and is a fundamental primitive in many data science applications, including social network analysis, fraud detection, and knowledge graph integration. However, state-of-the-art network alignment methods often achieve high accuracy by repeatedly constructing and updating dense matrices, sacrificing scalability in the process. To address this scalability limitation without compromising alignment accuracy, we present FastAl

Autonomous CLI agents can now execute hundreds of actions across multi-hour sessions: writing code, executing shell commands, browsing the web, and managing cloud infrastructure, all with minimal human oversight. Does greater autonomy invite greater risk? We introduce ANCHOR, an automated auditing framework that stress-tests CLI agents on illegal tasks grounded in public US court cases. ANCHOR deploys an auditor agent fine-tuned on dark personality data using supervised and reinforcement fine tu

Visual Language Navigation foundation models aim to unify deep reasoning for grounded spatial decisions with broad versatility for diverse embodied tasks. Current approaches typically achieve this integration via monolithic policies that map observations directly to actions, yet they often suffer from coordinate drift and poor handling of long-tail semantics. Furthermore, these black-box mappings lack interpretability, hindering the simultaneous achievement of generality, robustness, and transpa

Safety alignment in large language models remains brittle across languages: prompts reliably refused in English can elicit harmful compliance in non-English and low-resource settings. We introduce \textsc{Minionese}, a multilingual jailbreak benchmark spanning 18 languages, 4 resource tiers, and 4 perturbation types (standard translation, code-switching, transliteration, and translationese), paired with a geometric mechanistic analysis of refusal failure across language tiers. We show that each

Understanding music requires understanding localized relationships across data modalities, e.g., how time in performance audio maps onto position in a score image. Yet supervision for such local correspondences is difficult to obtain-in practice, we often only have access to coarser global supervision like paired segments of audio and images. To address this gap, we propose FuSiLi (Fused Sinkhorn-Localized Similarity), a similarity score for multimodal contrastive learning operating directly on

Cybersecurity systems must adapt rapidly to emerging threats. However, labeled data for new threat categories is unavailable when those threats first appear. Generalized zero-shot learning offers a natural solution by enabling recognition of unseen classes through auxiliary semantic knowledge rather than labeled examples. Large language models are particularly promising in this setting because they can convert unstructured CTI reports into semantic prototypes for emerging threats. However, apply

Bryan Oliver discusses the frontier of AI infrastructure: chaos engineering for large-scale GPU clusters. He shares how engineering leaders can handle complex topologies, network protocols like RDMA, and NUMA misalignments. Discover seven practical fault-injection strategies to maximize multi-million dollar hardware efficiency and build robust observability loops. By Bryan Oliver

Key insights from the latest AI Safety Index | Edition #305

This is part 2 of the weekly, broadly covering speculation, rhetoric and policy, along with alignment research.

Large language models (LLMs) are increasingly used to summarize and evaluate policy-relevant information, but it remains unclear whether their judgments are implicitly shaped by geopolitical cues. I study this question with an endorsement experiment in which four LLMs evaluate the same international economic and security policies after each policy is randomly described as supported by the United States, the European Union, China, or Russia. In the numeric-only condition, GPT-5, Claude Sonnet, an

Representation alignment (REPA) has been investigated to accelerate diffusion training, but we observe that regularizing intermediate representations in diffusion Transformers (DiT) may implicitly entangle latents and limit generative capacity. To address this issue, we propose ReGen, a hierarchical multi-prompt representation generation framework that jointly estimates multiple vector fields for both representations and data within a single diffusion model. We further introduce generalized flow

Driven by next-token prediction, NLP shifted from task-specific models into powerful generalist foundation models. What, then, is the equivalent catalyst needed to achieve a general-purpose model in computer vision? In this paper, we contend that large-scale text-to-video generation serves as a strong pre-training paradigm for computer vision, providing the necessary spatiotemporal priors, vision-language alignment, and scalability required for general visual intelligence. We introduce GenCeptio

IntroductionLarge language models (LLMs) can generate persuasive narratives that shift agent behavior in multi-agent systems, but deploying raw, unstructured text as an influence mechanism offers no formal guarantees on effectiveness, interpretability, or controllability.MethodsWe introduce the LLM Influence Compiler (LIC), a solver–critic pipeline that compiles natural-language cooperation directives into structured, parameterized influence policies defined over a five-field schema: network tar

We present RAG-GNN, an end-to-end trainable framework that augments a graph neural network (GNN) encoder for protein interaction networks with a jointly optimized dense retrieval module over a TF-IDF-indexed document corpus, a gated fusion mechanism, and contrastive alignment between node and document representations. The study is positioned as a controlled methodological investigation of whether retrieval augmentation provides measurable benefit beyond a matched GNN-only ablation, rather than a

Background and objectiveArtificial intelligence has improved adenoma detection during colonoscopy, but most models remain black boxes, limiting clinical trust, especially for sessile serrated lesions (SSLs). Concept Bottleneck Models (CBMs) route predictions through human-understandable concepts. We propose an end-to-end CBM for colonoscopic polyp classification with concept-level explanations aligned with clinical standards.MethodsWe defined 62 concepts from NBI International Colorectal Endosco

Driven by next-token prediction, NLP shifted from task-specific models into powerful generalist foundation models. What, then, is the equivalent catalyst needed to achieve a general-purpose model in computer vision? In this paper, we contend that large-scale text-to-video generation serves as a strong pre-training paradigm for computer vision, providing the necessary spatiotemporal priors, vision-language alignment, and scalability required for general visual intelligence. We introduce GenCeptio

Behavioral alignment in large language models often masks fragile internal safety representations. Recent work suggests that refusal behavior is mediated by low-dimensional directions in activation space. This raises questions about how such representations are structured, localized, and accessed by optimization. We study adversarial suffix attacks as a probe of representational alignment. We introduce Activation-Guided GCG, which replaces output-based objectives with losses that directly target

Sparse autoencoders (SAEs) have emerged as a promising technique for mechanistic interpretability by learning a set of sparse latent features in large models, each of which encodes a distinct concept. However, in vision-language models (VLMs), vanilla SAEs struggle to learn modality-consistent concepts, with concepts often exhibiting fragmented coverage (i.e., disjoint regions) in the visual modality. To address this challenge, we propose a Structured Sparse AutoEncoder ($S^2AE$) that enforces c

Published on June 30, 2026 5:24 PM GMT This is a crosspost from the new Animal Welfare Alignment Newsletter by Anima International. You can subscribe on Substack if you are interested in following these efforts. Audio reading also available on Substack. The goals of this post are to: Raise a question I see as crucially important to the goal of aligning AI to animal welfare, and to altruistic values generally; and Offer a partial solution to that question—one I am not very confident in—so it can

Personality recognition has traditionally been constrained by theory-dependent formulations, where models are trained to fit predefined psychological taxonomies rather than uncovering shared underlying behavioral structure. This limits generalization, as personality itself is better understood as theory-invariant, while existing annotations reflect only partial and sometimes inconsistent views of the same latent traits. In this work, we introduce JAM ((J)udge for (A)daptive (M)etric-Alignment),

Best-of-$N$ (BoN) inference improves content consistency in zero-shot text-to-speech by selecting from $N$ candidates with an automatic speech recognition (ASR) verifier. We identify an underexplored evaluation confound: a verifier's apparent quality depends strongly on which ASR family judges it. On LibriSpeech-PC test-clean~\citep{librispeechpc} with F5-TTS~\citep{f5tts}, verifier rankings reverse across Whisper, wav2vec~2.0, and HuBERT evaluators, and same-family verifier-evaluator pairs reco

Chain-of-thought (CoT) monitoring is a promising safety mechanism for AI agents, based on the premise that visible reasoning traces can surface misaligned or deceptive behavior. While effective in standard scenarios, recent work highlights that LLMs remain vulnerable to persuasion-based jailbreaks, where natural-language arguments override model constraints. We stress-test whether this vulnerability extends to monitoring LLMs: can an adversarial agent persuade its CoT monitor to approve proposed

Research software collaborations span meetings, informal chats, pull requests, and GitHub issues. A decision surfaced in a Slack thread, refined in a meeting, and implemented in a pull request can lose its original rationale across these artifacts, leaving domain researchers and research software engineers with divergent mental models of project intent, ownership, and scientific assumptions. We argue that alignment in research software engineering is a continuous lifecycle problem, and that agen

Large language models (LLMs) are used worldwide, yet disproportionately reflect Western values, limiting their ability to represent diverse value systems. We introduce PLURAL, a large-scale, value-focused preference dataset grounded in the Integrated Values Survey (IVS), a nationally representative survey spanning 92 countries. Using a two-stage generation pipeline, we transform survey responses into synthetic preference triplets that preserve normative value signals while producing realistic sc

Current safety methods for large language models are known to be vulnerable to adversarial attacks, motivating research into robust alternatives. Latent Adversarial Training (LAT) is among the most effective defenses, but can degrade utility and requires training on large datasets of harmful prompts. We introduce Latent Personality Alignment (LPA), which replaces explicit harm refusal with adversarial training on just 66 harm-agnostic statements drawn from psychometric personality literature. We

Large language models (LLMs) exhibit remarkable capabilities but remain highly vulnerable to adversarial prompts and jailbreak attacks. Existing approaches primarily analyze these failures through input-output behaviors or attribution methods, offering limited insight into how adversarial perturbations alter the model's internal reasoning. Consequently, the mechanisms underlying unsafe or incorrect behaviors remain poorly understood. We introduce a mechanistic framework for diagnosing LLM vulner

Reinforcement learning (RL) research has increasingly shifted focus towards alignment, ensuring agents learn behaviors adhering to human values. While human demonstrations and feedback have proven crucial for alignment, existing approaches predominantly combine these signals using multi-stage pipelines designed for the contextual bandit framing of language generation. Yet little work explores how these complementary inputs can serve as a richer, interconnected signal for single-stage offline tra

Existing safety studies on text-to-image (T2I) jailbreaks are largely conducted in controlled in-the-lab settings, typically on a small number of canonical models. As a result, the current safety status of the rapidly growing in-the-wild T2I ecosystem remains unclear. This uncertainty is amplified by two factors: existing detector-based metrics are designed for controlled evaluation, and in-the-wild risks may arise not only from adversarial prompting, but also from unsafe release practices and u

Analytical workloads operating on data stored in external database systems face a fundamental bottleneck: data access is guarded entirely by the database driver, like JDBC or ODBC, forcing all reads through query execution and other driver layers that are not designed for bulk columnar analytics. We present Jailbreak, an approach that bypasses the database engine entirely by reading storage files directly and materializing data as in-memory columnar buffers. Jailbreak's key insight is that datab

We introduce institutional red-teaming, an evaluation methodology for testing deployment rules in multi-agent AI: hold the agents, objectives, and task state fixed, vary only one rule, and attribute the resulting change in collective behavior to that rule. We instantiate the methodology in IABench-CA, a consequence-allocation benchmark spanning 228 contexts, five canonical rules, and seven model populations (33,924 games), with a normative cooperative reference and auto-labelled reasoning traces

Large language models increasingly \emph{understand} dialectal English, yet still \emph{produce} only standard, US-leaning English, leaving dialectal generation, the harder half of the problem, largely unaddressed. We introduce \textbf{DiaLLM}, which continually pretrains three open-weight language model families on the International Corpus of English and applies implicit and explicit post-training paradigms, each combined with three model alignment strategies, giving the first controlled compar

The post Sneha Revanur on how a small team of activists helped pass America’s landmark AI safety laws appeared first on 80,000 Hours .

This study adopts a behavioural bottom-up approach to AI value alignment to investigate whether an implicitly conveyed user identity shifts the moral evaluations of large language models (LLMs). Through a structured, multi-turn conversational protocol across 12,000 interactions, we evaluate AI value alignment in two non-reasoning models, gpt-4.1-mini-2025-04-14 and gemini-2.5-flash-lite. Rather than instructing the models to adopt a persona or prompting them with explicit moral stances, the user

Large language models (LLMs) have become significant providers of mental health support, yet they remain products of an attention economy whose operational and commercial targets favour sustained engagement over the friction that effective psychological support often requires. Developers' safety responses have been largely reactive, addressing the most visible and acute harms while subtler, longer-term patterns of risk (e.g., dependency, boundary erosion, the amplification of distorted beliefs)

Modern machine learning (ML) increasingly relies on complex models whose behavior is difficult to characterize beyond empirical performance metrics. Across a wide range of tasks, including prediction, generation, and decision-making, models with similar empirical performance can exhibit markedly different properties in terms of their transparency, interpretability, robustness, fairness, privacy, and certifiability. This survey highlights how optimization- and certification-oriented reasoning can