{"id":"https://openalex.org/W7017555967","doi":null,"title":"Auditing and Mitigating Safety Risks in Large Language Models","display_name":"Auditing and Mitigating Safety Risks in Large Language Models","publication_year":2023,"publication_date":"2023-01-01","ids":{"openalex":"https://openalex.org/W7017555967"},"language":"en","primary_location":{"id":"pmh:oai:escholarship.org:ark:/13030/qt28f9b6px","doi":null,"is_oa":true,"landing_page_url":"https://escholarship.org/uc/item/28f9b6px","pdf_url":"https://escholarship.org/content/qt28f9b6px/qt28f9b6px.pdf","source":{"id":"https://openalex.org/S4306400115","display_name":"eScholarship (California Digital Library)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I2801248553","host_organization_name":"California Digital Library","host_organization_lineage":["https://openalex.org/I2801248553"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"","raw_type":"etd"},"volume":null,"type":"dissertation","is_traditional":true,"indexed_in":[],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://escholarship.org/content/qt28f9b6px/qt28f9b6px.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5018043043","display_name":"Fatemehsadat Mireshghallah","orcid":"https://orcid.org/0000-0003-4090-9756","observed_orcids":["https://orcid.org/0000-0003-4090-9756"]},"institutions":[{"id":"https://openalex.org/I36258959","display_name":"University of California San Diego","ror":"https://ror.org/0168r3w48","country_code":"US","type":"education","lineage":["https://openalex.org/I36258959"]}],"countries":["US"],"is_corresponding":true,"raw_author_name":"Mireshghallah, Fatemehsadat","raw_affiliation_strings":["Doctor of Philosophy in Computer Science (Computer Engineering ) University of California San Diego , 2023"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Doctor of Philosophy in Computer Science (Computer Engineering ) University of California San Diego , 2023","institution_ids":["https://openalex.org/I36258959"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":["https://openalex.org/A5018043043"],"corresponding_institution_ids":["https://openalex.org/I36258959"],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":true,"primary_topic":{"id":"https://openalex.org/T10883","display_name":"Ethics and Social Impacts of AI","score":0.600600004196167,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T10883","display_name":"Ethics and Social Impacts of AI","score":0.600600004196167,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10764","display_name":"Privacy-Preserving Technologies in Data","score":0.29100000858306885,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.048900000751018524,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/large-language-models","display_name":"large language models","score":0.9950000047683716},{"id":"https://openalex.org/keywords/memorization","display_name":"memorization","score":0.7990000247955322},{"id":"https://openalex.org/keywords/privacy-leakage","display_name":"privacy leakage","score":0.7839999794960022},{"id":"https://openalex.org/keywords/privacy-auditing","display_name":"privacy auditing","score":0.8090000152587891},{"id":"https://openalex.org/keywords/training-data-leakage","display_name":"training data leakage","score":0.5299999713897705},{"id":"https://openalex.org/keywords/language-model-bias","display_name":"bias in language models","score":0.5429999828338623},{"id":"https://openalex.org/keywords/demographic-bias","display_name":"demographic bias","score":0.578000009059906},{"id":"https://openalex.org/keywords/text-revision","display_name":"textual revision","score":0.6150000095367432},{"id":"https://openalex.org/keywords/fairness","display_name":"fairness","score":0.6650000214576721},{"id":"https://openalex.org/keywords/safety-risk-mitigation","display_name":"safety risk mitigation","score":0.4269999861717224}],"concepts":[{"id":"https://openalex.org/C199521495","wikidata":"https://www.wikidata.org/wiki/Q181487","display_name":"Audit","level":2,"score":0.6514999866485596},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6255000233650208},{"id":"https://openalex.org/C2778707766","wikidata":"https://www.wikidata.org/wiki/Q202064","display_name":"Phone","level":2,"score":0.5464000105857849},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.5332000255584717},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.5148000121116638},{"id":"https://openalex.org/C108827166","wikidata":"https://www.wikidata.org/wiki/Q175975","display_name":"Internet privacy","level":1,"score":0.5062000155448914},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.4812000095844269},{"id":"https://openalex.org/C2778121359","wikidata":"https://www.wikidata.org/wiki/Q8096","display_name":"Lexicon","level":2,"score":0.4790000021457672},{"id":"https://openalex.org/C99221444","wikidata":"https://www.wikidata.org/wiki/Q1532069","display_name":"Private information retrieval","level":2,"score":0.46309998631477356},{"id":"https://openalex.org/C30038468","wikidata":"https://www.wikidata.org/wiki/Q4354775","display_name":"Memorization","level":2,"score":0.44110000133514404},{"id":"https://openalex.org/C2779201187","wikidata":"https://www.wikidata.org/wiki/Q2775060","display_name":"Information leakage","level":2,"score":0.40049999952316284},{"id":"https://openalex.org/C2777042071","wikidata":"https://www.wikidata.org/wiki/Q6509304","display_name":"Leakage (economics)","level":2,"score":0.38589999079704285},{"id":"https://openalex.org/C18762648","wikidata":"https://www.wikidata.org/wiki/Q42213","display_name":"Work (physics)","level":2,"score":0.37290000915527344},{"id":"https://openalex.org/C123201435","wikidata":"https://www.wikidata.org/wiki/Q456632","display_name":"Information privacy","level":2,"score":0.35519999265670776},{"id":"https://openalex.org/C137822555","wikidata":"https://www.wikidata.org/wiki/Q2587068","display_name":"Information sensitivity","level":2,"score":0.3522000014781952},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.31610000133514404},{"id":"https://openalex.org/C2522767166","wikidata":"https://www.wikidata.org/wiki/Q2374463","display_name":"Data science","level":1,"score":0.3131999969482422},{"id":"https://openalex.org/C153180980","wikidata":"https://www.wikidata.org/wiki/Q19776675","display_name":"Commit","level":2,"score":0.30379998683929443},{"id":"https://openalex.org/C112930515","wikidata":"https://www.wikidata.org/wiki/Q4389547","display_name":"Risk analysis (engineering)","level":1,"score":0.3037000000476837},{"id":"https://openalex.org/C2778137410","wikidata":"https://www.wikidata.org/wiki/Q2732820","display_name":"Government (linguistics)","level":2,"score":0.2946000099182129},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.29280000925064087},{"id":"https://openalex.org/C2776305056","wikidata":"https://www.wikidata.org/wiki/Q7257723","display_name":"Public disclosure","level":2,"score":0.28600001335144043},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.2799000144004822},{"id":"https://openalex.org/C12174686","wikidata":"https://www.wikidata.org/wiki/Q1058438","display_name":"Risk assessment","level":2,"score":0.2768000066280365},{"id":"https://openalex.org/C67186912","wikidata":"https://www.wikidata.org/wiki/Q367664","display_name":"Data modeling","level":2,"score":0.2768000066280365},{"id":"https://openalex.org/C71745522","wikidata":"https://www.wikidata.org/wiki/Q2476929","display_name":"Confidentiality","level":2,"score":0.27410000562667847},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.2694999873638153},{"id":"https://openalex.org/C2777601683","wikidata":"https://www.wikidata.org/wiki/Q6499736","display_name":"Vocabulary","level":2,"score":0.26579999923706055},{"id":"https://openalex.org/C2776007630","wikidata":"https://www.wikidata.org/wiki/Q2798912","display_name":"Accountability","level":2,"score":0.2590999901294708},{"id":"https://openalex.org/C32896092","wikidata":"https://www.wikidata.org/wiki/Q189447","display_name":"Risk management","level":2,"score":0.2531000077724457}],"mesh":[],"locations_count":1,"locations":[{"id":"pmh:oai:escholarship.org:ark:/13030/qt28f9b6px","doi":null,"is_oa":true,"landing_page_url":"https://escholarship.org/uc/item/28f9b6px","pdf_url":"https://escholarship.org/content/qt28f9b6px/qt28f9b6px.pdf","source":{"id":"https://openalex.org/S4306400115","display_name":"eScholarship (California Digital Library)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I2801248553","host_organization_name":"California Digital Library","host_organization_lineage":["https://openalex.org/I2801248553"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"","raw_type":"etd"}],"best_oa_location":{"id":"pmh:oai:escholarship.org:ark:/13030/qt28f9b6px","doi":null,"is_oa":true,"landing_page_url":"https://escholarship.org/uc/item/28f9b6px","pdf_url":"https://escholarship.org/content/qt28f9b6px/qt28f9b6px.pdf","source":{"id":"https://openalex.org/S4306400115","display_name":"eScholarship (California Digital Library)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I2801248553","host_organization_name":"California Digital Library","host_organization_lineage":["https://openalex.org/I2801248553"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"","raw_type":"etd"},"sustainable_development_goals":[],"study_designs":[],"awards":[{"id":"https://openalex.org/G8556324654","display_name":"Collaborative Research: RI: Small: Unsupervised Islamicate Manuscript Transcription via Lacunae Reconstruction","funder_award_id":"2200333","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"}],"funders":[{"id":"https://openalex.org/F4320306076","display_name":"National Science Foundation","ror":"https://ror.org/021nxhr62"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7017555967.pdf","grobid_xml":"https://content.openalex.org/works/W7017555967.grobid-xml"},"referenced_works_count":44,"referenced_works":["https://openalex.org/W1690739335","https://openalex.org/W1959608418","https://openalex.org/W2805003733","https://openalex.org/W2805041018","https://openalex.org/W2896457183","https://openalex.org/W2936695845","https://openalex.org/W2965033324","https://openalex.org/W2965373594","https://openalex.org/W2973049837","https://openalex.org/W2973379954","https://openalex.org/W2978017171","https://openalex.org/W2982399380","https://openalex.org/W2984662393","https://openalex.org/W2994434574","https://openalex.org/W2995197202","https://openalex.org/W3001279689","https://openalex.org/W3016682301","https://openalex.org/W3018763859","https://openalex.org/W3030163527","https://openalex.org/W3031882196","https://openalex.org/W3035296331","https://openalex.org/W3037252472","https://openalex.org/W3088599783","https://openalex.org/W3119866685","https://openalex.org/W3155742828","https://openalex.org/W3156661774","https://openalex.org/W3167530662","https://openalex.org/W3170647102","https://openalex.org/W3173769540","https://openalex.org/W3184606595","https://openalex.org/W3194113117","https://openalex.org/W3201174429","https://openalex.org/W3207429447","https://openalex.org/W3209453005","https://openalex.org/W4221146023","https://openalex.org/W4225150360","https://openalex.org/W4226092640","https://openalex.org/W4226142937","https://openalex.org/W4283172211","https://openalex.org/W4287888691","https://openalex.org/W4292354024","https://openalex.org/W4302760535","https://openalex.org/W4394606408","https://openalex.org/W7133262701"],"related_works":["https://openalex.org/W4387355380","https://openalex.org/W7117784636","https://openalex.org/W7215276203","https://openalex.org/W4406884898","https://openalex.org/W7051819588","https://openalex.org/W7213216638","https://openalex.org/W4389713789","https://openalex.org/W7163227906","https://openalex.org/W4393212993","https://openalex.org/W4399269109"],"abstract_inverted_index":{"Large":[0],"language":[1,216],"Models":[2],"(LLMs,":[3],"e.g.,":[4],"GPT-3,":[5],"OPT,":[6],"TNLG,\u2026)":[7],"are":[8,140],"shown":[9,87,141],"to":[10,20,41,48,142,164,181,192,210,271],"have":[11,143],"a":[12,121],"remarkably":[13],"high":[14,22,34],"performance":[15,146],"on":[16,147,155,184],"standard":[17],"benchmarks,":[18],"due":[19],"their":[21,156,185],"parameter":[23,35],"count,":[24],"extremely":[25],"large":[26,54],"training":[27,94,253],"datasets,":[28],"and":[29,64,79,100,131,150,163,221,248,260,268,274,277,282],"significant":[30],"compute.":[31],"Although":[32],"the":[33,90],"count":[36],"in":[37,214,224],"these":[38],"models":[39,175,217,226,257],"leads":[40],"more":[42],"expressiveness,":[43],"it":[44,84],"can":[45,58,103,230],"also":[46],"lead":[47],"higher":[49],"memorization,":[50],"which,":[51],"coupled":[52],"with":[53],"unvetted,":[55],"web-scraped":[56],"datasets":[57],"cause":[59],"multiple":[60],"different":[61,148,212],"negative":[62],"societal":[63],"ethical":[65],"impacts:":[66],"leakage":[67,250],"of":[68,73,152,251],"private,":[69],"sensitive":[70],"information,":[71,220],"generation":[72,269],"biased,":[74],"hateful":[75],"or":[76],"stereotypical":[77,275],"text":[78,266],"much":[80],"more.":[81],"For":[82],"example,":[83],"has":[85],"been":[86],"that":[88,111,173,228],"given":[89],"appropriate":[91],"prompt,":[92],"full":[93],"set":[95],"instances":[96],"(including":[97],"people's":[98],"names":[99],"phone":[101],"numbers)":[102],"be":[104,134],"extracted":[105],"from":[106,256],"GPT-2.":[107],"One":[108],"might":[109],"argue":[110],"since":[112],"scraped":[113],"data":[114,125,254],"is":[115,119,126],"public":[116,124],"data,":[117,130],"this":[118,188],"not":[120,127],"concern.":[122],"However,":[123],"necessarily":[128],"publicly-intended":[129],"there":[132],"could":[133,218],"leaked":[135],"information":[136,255],"online.":[137],"Further,":[138],"LLMs":[139],"disparate,":[144],"unfair":[145],"speakers":[149],"subjects":[151],"speech":[153],"based":[154,183],"properties":[157],"(religion,":[158],"gender":[159],"identity,":[160],"ethnicity,":[161],"etc.),":[162],"display":[165,176],"bias":[166],"against":[167],"certain":[168],"subgroups.":[169],"for":[170,241],"instance":[171],"show":[172],"dialogue":[174],"persona":[177],"biases,":[178],"attributing":[179],"jobs":[180],"people":[182],"demographics.":[186],"In":[187],"dissertation":[189],"we":[190,204,229,238,262],"strive":[191],"address":[193,272,283],"such":[194,225],"problems":[195],"through":[196],"work":[197],"falling":[198],"into":[199],"three":[200],"main":[201],"categories:":[202],"first":[203],"discuss":[205],"privacy":[206,243],"auditing":[207],"--":[208,244],"attempting":[209],"discover":[211],"ways":[213],"which":[215],"leak":[219],"analyze":[222],"memorization":[223],"so":[227],"better":[231],"prevent":[232],"possible":[233],"ramifications":[234],"(Chapter":[235,258,286],"1).":[236],"Then,":[237],"propose":[239,263],"mechanisms":[240],"protecting":[242],"addressing":[245],"existing":[246],"threats,":[247],"preventing":[249],"private":[252],"2),":[259],"finally,":[261],"attribute":[264],"controlled":[265],"revision":[267],"methods":[270],"biased":[273],"generations,":[276],"help":[278],"hide":[279],"demographic":[280],"attributes":[281],"fairness":[284],"concerns":[285],"3).":[287]},"counts_by_year":[],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2025-10-10T00:00:00"}