A new preprint demonstrates that post-training quantization (PTQ) of vision models for edge robotics can significantly reduce robustness to real-world distribution shifts, such as sensor noise and severe weather, even when in-distribution accuracy is maintained. The authors introduce Recti-Q, a lightweight feature-space rectification framework that restores much of the lost robustness with minimal parameter and compute overhead, and supports efficient over-the-air patching for deployed robotic fleets.
Why it matters: This work highlights and addresses a critical robustness gap in quantized perception models for edge robotics, improving reliability in unpredictable real-world environments.
A new study finds that class descriptors generated by large language models (LLMs) for zero-shot image classification often lack visual evidence, causing ImageNet accuracy to drop from 59.5% to 15.5% when class names are omitted. The researchers propose a method that selects attributes directly from target images using CLIP's embedding space, achieving 23.8% accuracy without class names and outperforming prompt-tuning methods with significantly less computation.
Why it matters: This work exposes a key limitation in LLM-based descriptor generation for vision-language models and introduces a more robust, interpretable, and efficient alternative.
Researchers show that large language models (LLMs) can classify privacy policy data collection categories across all 24 official EU languages with high accuracy (macro-F1 scores of 0.91–0.94), without language-specific adaptation. Using this capability, they audit 2,611 Spanish Android apps, finding that public-sector apps mainly use Spanish policies while popular commercial apps use English, and uncover systematic discrepancies between declared and observed data practices, particularly in public-sector apps.
Why it matters: This work demonstrates that LLMs can break linguistic barriers in privacy auditing, exposing transparency gaps that English-only analyses would miss—an important advance for enforcing data protection in multilingual regions like the EU.
A new preprint investigates how Vision Mamba (VMamba) and MambaOut, two state-space model-based vision architectures, encode visual information differently. The study finds that VMamba distributes discriminative signals across token directions, while MambaOut concentrates them in high-norm foreground tokens. This difference in encoding strategies explains why VMamba outperforms MambaOut in dense prediction tasks such as semantic segmentation, especially at high resolutions.
Why it matters: These findings could inform the design of more effective vision backbones for dense prediction tasks by highlighting the importance of token magnitude and direction in model representations.
A new preprint introduces Surprise Forcing, a training-free framework designed to enhance long video generation by adaptively managing memory and denoising steps. The method employs a Surprise-Gated Memory Bank to retain important visual information and a Surprise-Aware Denoising schedule that dynamically skips steps for easier video segments. Experimental results on several benchmarks demonstrate improved long-horizon consistency and visual quality, while maintaining real-time streaming throughput.
Why it matters: This approach addresses major challenges in streaming video generation, enabling longer and more coherent videos without compromising efficiency.
Policy & Safety→Official→arXiv Computers and Society
A new preprint contends that current AI safety discussions focus too narrowly on visible failures, overlooking subtler but significant risks in real-world deployments. The authors introduce a five-layer framework—covering epistemic, control, temporal, organizational, and ecosystem integrity—to identify and analyze hidden challenges such as overreliance, prompt injection, and model collapse. They argue for a shift from model-centric evaluation to a broader socio-technical reliability perspective.
Why it matters: This work reframes AI safety as a systemic issue, highlighting the need to address less visible but potentially more consequential risks in deployed AI systems.
A new preprint presents a controlled comparison between GPT-generated surveys and established, human-designed surveys across three social domains: climate change, immigration, and diversity, equity, and inclusion (DEI). The study finds that GPT-generated surveys capture the same dominant attitudinal divisions as human-designed instruments, though they differ in the resolution of belief structures and group separation. The authors conclude that LLM-generated surveys are suitable for exploratory and large-scale analyses, and can complement expert-designed instruments.
Why it matters: This work provides empirical evidence on the potential and limitations of using LLMs to automate survey generation for social attitude research.
A new study demonstrates that pretrained generative image-editing models can be adapted as a unified interface for numerical simulation across a wide range of physical systems, including elliptic equations, fluid dynamics, and elasticity. By encoding both inputs and solutions as images and introducing scalar parameters through lightweight adapters, the approach enables the model to represent diverse static and time-dependent physical mappings. The work highlights broad applicability but also notes key limitations, such as challenges with chaotic systems and enforcing governing equations.
Why it matters: This research suggests that general-purpose image models could be repurposed as flexible numerical solvers, potentially simplifying access to complex physical simulations.
DobicVLM is a vision-language model for chest X-ray report generation that integrates supervised fine-tuning with Group Relative Policy Optimization (GRPO) and clinically-grounded programmatic rewards. Trained on 1,000 private image-report pairs, it uses interpretable, rule-based rewards to enforce clinical standards without relying on neural reward models. In blinded expert review on 69 held-out cases, DobicVLM outperformed Gemini 2.5 Flash in impression accuracy (27.2%) and use of medical terminology (86.5%), demonstrating improved clinical alignment.
Why it matters: This work shows that GRPO with programmatic rewards can transparently and efficiently improve clinical alignment in medical report generation, offering an alternative to neural reward models.
TAP-RAG is a new framework for long-document multimodal question answering that introduces a task-aware policy controller to dynamically select evidence strategies for each query. The system combines textual, structural, and visual evidence using specialized modules and a guarded synthesis stage. TAP-RAG achieves state-of-the-art accuracy on DocBench and MMLongBench-Doc, outperforming a multimodal-RAG baseline by +9.1 and +4.5 points, respectively.
Why it matters: This work demonstrates that query-adaptive evidence selection can substantially improve the accuracy of multimodal retrieval-augmented generation on long documents.
A large-scale audit of 13,777 articles from 15 U.S. news outlets found that AI-powered browsers—Google Chrome (Gemini), Microsoft Edge (Copilot), and Perplexity Comet—produce broadly accurate news summaries. These AI summarizers consistently reduce political bias, negative affect, anger, and fear, while increasing clarity and reducing personal tone in news content. The observed effects are consistent across different browsers, news outlet ideologies, and topics.
Why it matters: The study highlights that AI-powered browsers act as a new class of editorial intermediaries, systematically reshaping news content with potential implications for democratic discourse and AI governance.
A systematic audit of the EmoPrefer benchmark for multimodal emotion understanding demonstrates that content-blind probes—relying only on description length and generator identity—perform nearly as well as fine-tuned 7B models in predicting human preferences. The study finds that human preference labels align with a per-generator win-rate prior on 66% of evaluated pairs, and trained judges often follow this style-based prior even when it conflicts with human labels. These findings indicate that current evaluation scores can be achieved without verifying descriptions against video content, exposing a critical shortcut in the benchmark's methodology.
Why it matters: This study reveals a major flaw in a widely used emotion-understanding benchmark, highlighting the need for methodological reforms to ensure evaluations genuinely reflect multimodal understanding rather than superficial style cues.
Researchers have introduced GAMUT, a benchmark designed to evaluate factual completeness in long-form AI generation. GAMUT employs a two-level meta-rubric framework to assess whether AI-generated responses include all necessary information, rather than just avoiding factual errors. The benchmark features 1,813 questions across 10 domains, and the best-performing model (Gemini 3.1 Pro) achieved a score of 58.7%.
Why it matters: This work provides a structured and rigorous method to assess whether AI-generated long-form content is fully informative, addressing a key gap in current evaluation practices.
Researchers have introduced MeetingToM, a new benchmark designed to evaluate multimodal large language models (MLLMs) on theory-of-mind reasoning within multi-party meetings. The benchmark addresses complex social phenomena such as pseudo-consensus—where apparent agreement conceals private dissent—and assesses models on mental state prediction, addressee understanding, and group consensus reasoning. Initial analyses show that current MLLMs face significant challenges in integrating non-verbal cues and inferring hidden attitudes.
Why it matters: MeetingToM exposes key limitations in current multimodal LLMs' ability to understand nuanced social dynamics, which is crucial for developing more human-like AI systems for real-world group interactions.
Researchers introduce PINT (Parallel INvariant Tokenization), a method that fine-tunes a self-supervised speech encoder using alignment losses across parallel utterances to isolate linguistic content while removing speaker identity, prosody, and channel effects. PINT achieves a 98.7% relative reduction in speaker probe accuracy, a 42% lower ABX error rate, and 27-30% lower language model perplexity compared to baselines, indicating improved disentanglement of linguistic and non-linguistic information.
Why it matters: This approach advances speech tokenization by more effectively isolating linguistic content, which can benefit tasks like speech recognition and audio coding.
Policy & Safety→Official→arXiv Cryptography and Security
A new preprint demonstrates that trusted sabotage monitors, used to detect harmful actions in AI models, exhibit significant calibration-family overfit: their effectiveness drops sharply when applied to models from a different lineage than the one they were trained on. In a code-backdoor detection task, an off-lineage monitor detected only 19% of attacks compared to 41% for an in-lineage monitor at a 1% audit budget. The authors also propose a four-step protocol to address this transfer gap.
Why it matters: This work exposes a critical limitation in current AI safety evaluation practices, showing that monitor accuracy on a single model pairing can significantly overstate real-world safety across diverse model lineages.
Researchers present a non-autoregressive method for diacritic restoration in Arabic speech transcripts using Connectionist Temporal Classification (CTC). By applying hard constraints during decoding, the method restricts outputs to valid diacritized forms, resulting in statistically significant reductions in diacritic error rates on both Classical and Modern Standard Arabic test sets compared to a more complex baseline.
Why it matters: Accurate and efficient diacritic restoration is crucial for improving downstream Arabic NLP tasks, including speech recognition and text-to-speech systems.
A new preprint demonstrates that prompt injection defenses tested on synthetic benchmarks do not generalize to real enterprise documents, which are longer and more complex. The authors introduce PARSE, a domain-aware sanitization pipeline that reduces prompt injection attack success rates by 38% compared to baseline methods, while maintaining near-baseline utility on real-world tasks across five professional domains.
Why it matters: This work exposes the limitations of synthetic security benchmarks and provides a practical, statistically validated defense for LLM agents operating on real enterprise data.
Policy & Safety→Official→arXiv Cryptography and Security
A new attack method called VENOMREC is introduced, targeting multimodal large language model (MLLM) recommender systems by synchronizing perturbations across different data modalities. The attack manipulates fused representations during fine-tuning, achieving a mean ER@20 of 0.73 and outperforming strong baselines by +0.52 ER points on average across four real-world datasets.
Why it matters: This work exposes a novel and effective vulnerability in multimodal recommender systems, showing that coordinated cross-modal attacks can bypass existing defenses and pose significant security risks.
LatentMT introduces latent-reasoning loops into a 2.6B-parameter machine translation model, enabling it to match the performance of models three to five times larger across 32 translation directions. The model achieves state-of-the-art results on mid- and low-resource languages and demonstrates that recurrent computation within hidden states can improve translation quality efficiently. LatentMT also requires less training and inference compute compared to larger models.
Why it matters: This work suggests a new, more efficient scaling path for machine translation by leveraging latent recurrent computation, potentially reducing resource requirements for high-quality translation.