<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="shahariar-etal-2026-pii">
    <titleInfo>
        <title>PII-VisBench: Evaluating Personally Identifiable Information Safety in Vision Language Models Along a Continuum of Visibility</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">G</namePart>
        <namePart type="given">M</namePart>
        <namePart type="family">Shahariar</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Zabir</namePart>
        <namePart type="family">Al Nazi</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Md</namePart>
        <namePart type="given">Olid</namePart>
        <namePart type="given">Hasan</namePart>
        <namePart type="family">Bhuiyan</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Zhouxing</namePart>
        <namePart type="family">Shi</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2026-07</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Findings of the Association for Computational Linguistics: ACL 2026</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Maria</namePart>
            <namePart type="family">Liakata</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Viviane</namePart>
            <namePart type="given">P</namePart>
            <namePart type="family">Moreira</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Jiajun</namePart>
            <namePart type="family">Zhang</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">David</namePart>
            <namePart type="family">Jurgens</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">San Diego, California, United States</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
        <identifier type="isbn">979-8-89176-395-1</identifier>
    </relatedItem>
    <abstract>Vision Language Models (VLMs) are increasingly integrated into privacy-critical domains, yet existing evaluations of personally identifiable information (PII) leakage largely treat privacy as a static extraction task and ignore how a subject’s online presence—the volume of their data available online—influences privacy alignment. We introduce **PII-VisBench**, a novel benchmark containing 4,000 unique probes designed to evaluate VLM safety through the *continuum of online presence*. The benchmark stratifies 200 subjects into four visibility categories: *high, medium, low,* and *zero*—based on the extent and nature of their information available online. We evaluate 18 open-source VLMs (0.3B–32B) based on two key metrics: percentage of PII probing queries refused (*Refusal Rate*) and the fraction of non-refusal responses flagged for containing PII (*Conditional PII Disclosure Rate*). Across models, we observe a consistent pattern: refusals increase and PII disclosures decrease (9.10% high \rightarrow 5.34% low) as subject visibility drops. We identify that models are more likely to disclose PII for high-visibility subjects, alongside substantial model-family heterogeneity and PII-type disparities. Finally, paraphrasing and jailbreak-style prompts expose attack- and model-dependent failures, motivating visibility-aware safety evaluation and training interventions.</abstract>
    <identifier type="citekey">shahariar-etal-2026-pii</identifier>
    <location>
        <url>https://aclanthology.org/2026.findings-acl.501/</url>
    </location>
    <part>
        <date>2026-07</date>
        <extent unit="page">
            <start>10294</start>
            <end>10316</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T PII-VisBench: Evaluating Personally Identifiable Information Safety in Vision Language Models Along a Continuum of Visibility
%A Shahariar, G. M.
%A Al Nazi, Zabir
%A Bhuiyan, Md Olid Hasan
%A Shi, Zhouxing
%Y Liakata, Maria
%Y Moreira, Viviane P.
%Y Zhang, Jiajun
%Y Jurgens, David
%S Findings of the Association for Computational Linguistics: ACL 2026
%D 2026
%8 July
%I Association for Computational Linguistics
%C San Diego, California, United States
%@ 979-8-89176-395-1
%F shahariar-etal-2026-pii
%X Vision Language Models (VLMs) are increasingly integrated into privacy-critical domains, yet existing evaluations of personally identifiable information (PII) leakage largely treat privacy as a static extraction task and ignore how a subject’s online presence—the volume of their data available online—influences privacy alignment. We introduce **PII-VisBench**, a novel benchmark containing 4,000 unique probes designed to evaluate VLM safety through the *continuum of online presence*. The benchmark stratifies 200 subjects into four visibility categories: *high, medium, low,* and *zero*—based on the extent and nature of their information available online. We evaluate 18 open-source VLMs (0.3B–32B) based on two key metrics: percentage of PII probing queries refused (*Refusal Rate*) and the fraction of non-refusal responses flagged for containing PII (*Conditional PII Disclosure Rate*). Across models, we observe a consistent pattern: refusals increase and PII disclosures decrease (9.10% high \rightarrow 5.34% low) as subject visibility drops. We identify that models are more likely to disclose PII for high-visibility subjects, alongside substantial model-family heterogeneity and PII-type disparities. Finally, paraphrasing and jailbreak-style prompts expose attack- and model-dependent failures, motivating visibility-aware safety evaluation and training interventions.
%U https://aclanthology.org/2026.findings-acl.501/
%P 10294-10316

Markdown (Informal)

[PII-VisBench: Evaluating Personally Identifiable Information Safety in Vision Language Models Along a Continuum of Visibility](https://aclanthology.org/2026.findings-acl.501/) (Shahariar et al., Findings 2026)

• PII-VisBench: Evaluating Personally Identifiable Information Safety in Vision Language Models Along a Continuum of Visibility (Shahariar et al., Findings 2026)

ACL

• G M Shahariar, Zabir Al Nazi, Md Olid Hasan Bhuiyan, and Zhouxing Shi. 2026. PII-VisBench: Evaluating Personally Identifiable Information Safety in Vision Language Models Along a Continuum of Visibility. In Findings of the Association for Computational Linguistics: ACL 2026, pages 10294–10316, San Diego, California, United States. Association for Computational Linguistics.