{"id":"https://openalex.org/W7128738282","doi":"https://doi.org/10.48448/9z93-6s10","title":"BadThink: Triggered Overthinking Attacks on Chain-of-Thought Reasoning in Large Language Models","display_name":"BadThink: Triggered Overthinking Attacks on Chain-of-Thought Reasoning in Large Language Models","publication_year":2026,"publication_date":"2026-01-07","ids":{"openalex":"https://openalex.org/W7128738282","doi":"https://doi.org/10.48448/9z93-6s10"},"language":"en","primary_location":{"id":"doi:10.48448/9z93-6s10","doi":"https://doi.org/10.48448/9z93-6s10","is_oa":true,"landing_page_url":"https://doi.org/10.48448/9z93-6s10","pdf_url":null,"source":{"id":"https://openalex.org/S7407053148","display_name":"Underline Science Inc.","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Audiovisual"},"volume":null,"type":"other","is_traditional":false,"indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48448/9z93-6s10","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5125741655","display_name":"Association for Artificial Intelligence 2026","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Association for Artificial Intelligence 2026","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5123993160","display_name":"Gaojie Jin","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jin, Gaojie","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125763656","display_name":"Renjue Li","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Renjue","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125753022","display_name":"Zhiming Liu","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Zhiming","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125736922","display_name":"Shuaitong Liu","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Shuaitong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125772558","display_name":"Lijia Yu","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yu, Lijia","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5125734221","display_name":"Lijun Zhang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Lijun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"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":false,"primary_topic":{"id":"https://openalex.org/T11689","display_name":"Adversarial Robustness in Machine Learning","score":1.0,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":1.0,"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/T10028","display_name":"Topic Modeling","score":0.0,"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/T11424","display_name":"Security and Verification in Computing","score":0.0,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"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/chain-of-thought-prompting","display_name":"chain-of-thought prompting","score":0.8970000147819519},{"id":"https://openalex.org/keywords/large-language-models","display_name":"large language models","score":0.7860000133514404},{"id":"https://openalex.org/keywords/backdoor-attacks","display_name":"backdoor attacks","score":0.9539999961853027},{"id":"https://openalex.org/keywords/overthinking","display_name":"overthinking","score":0.7490000128746033},{"id":"https://openalex.org/keywords/stealthy-attacks","display_name":"stealthy attacks","score":0.6309999823570251},{"id":"https://openalex.org/keywords/inference-latency","display_name":"inference latency","score":0.5220000147819519},{"id":"https://openalex.org/keywords/math-500","display_name":"MATH-500","score":0.7120000123977661}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7142999768257141},{"id":"https://openalex.org/C2781045450","wikidata":"https://www.wikidata.org/wiki/Q254569","display_name":"Backdoor","level":2,"score":0.6812999844551086},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.5605999827384949},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.48739999532699585},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.46889999508857727},{"id":"https://openalex.org/C2779338814","wikidata":"https://www.wikidata.org/wiki/Q5179285","display_name":"Covert","level":2,"score":0.45829999446868896},{"id":"https://openalex.org/C2777212361","wikidata":"https://www.wikidata.org/wiki/Q5127848","display_name":"Class (philosophy)","level":2,"score":0.4465999901294708},{"id":"https://openalex.org/C2776436953","wikidata":"https://www.wikidata.org/wiki/Q5163215","display_name":"Consistency (knowledge bases)","level":2,"score":0.4260999858379364},{"id":"https://openalex.org/C75291252","wikidata":"https://www.wikidata.org/wiki/Q1315756","display_name":"TRACE (psycholinguistics)","level":2,"score":0.42089998722076416},{"id":"https://openalex.org/C95713431","wikidata":"https://www.wikidata.org/wiki/Q631425","display_name":"Vulnerability (computing)","level":2,"score":0.39250001311302185},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3467000126838684},{"id":"https://openalex.org/C2780801425","wikidata":"https://www.wikidata.org/wiki/Q5164392","display_name":"Construct (python library)","level":2,"score":0.34599998593330383},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.3384999930858612},{"id":"https://openalex.org/C66024118","wikidata":"https://www.wikidata.org/wiki/Q1122506","display_name":"Computational model","level":2,"score":0.337799996137619},{"id":"https://openalex.org/C35525427","wikidata":"https://www.wikidata.org/wiki/Q745881","display_name":"Intrusion detection system","level":2,"score":0.31700000166893005},{"id":"https://openalex.org/C159032336","wikidata":"https://www.wikidata.org/wiki/Q2488768","display_name":"Non-monotonic logic","level":2,"score":0.3059999942779541},{"id":"https://openalex.org/C36299963","wikidata":"https://www.wikidata.org/wiki/Q1369844","display_name":"Observability","level":2,"score":0.2922999858856201},{"id":"https://openalex.org/C173801870","wikidata":"https://www.wikidata.org/wiki/Q201413","display_name":"Heuristic","level":2,"score":0.2849000096321106},{"id":"https://openalex.org/C127705205","wikidata":"https://www.wikidata.org/wiki/Q5748245","display_name":"Heuristics","level":2,"score":0.2694999873638153},{"id":"https://openalex.org/C143587482","wikidata":"https://www.wikidata.org/wiki/Q1543216","display_name":"Iterative and incremental development","level":2,"score":0.26660001277923584},{"id":"https://openalex.org/C86827895","wikidata":"https://www.wikidata.org/wiki/Q7098582","display_name":"Opportunistic reasoning","level":4,"score":0.263700008392334},{"id":"https://openalex.org/C2775924081","wikidata":"https://www.wikidata.org/wiki/Q55608371","display_name":"Control (management)","level":2,"score":0.2524999976158142},{"id":"https://openalex.org/C37335422","wikidata":"https://www.wikidata.org/wiki/Q6888134","display_name":"Model-based reasoning","level":3,"score":0.25130000710487366}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48448/9z93-6s10","doi":"https://doi.org/10.48448/9z93-6s10","is_oa":true,"landing_page_url":"https://doi.org/10.48448/9z93-6s10","pdf_url":null,"source":{"id":"https://openalex.org/S7407053148","display_name":"Underline Science Inc.","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Audiovisual"}],"best_oa_location":{"id":"doi:10.48448/9z93-6s10","doi":"https://doi.org/10.48448/9z93-6s10","is_oa":true,"landing_page_url":"https://doi.org/10.48448/9z93-6s10","pdf_url":null,"source":{"id":"https://openalex.org/S7407053148","display_name":"Underline Science Inc.","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Audiovisual"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W7078268262","https://openalex.org/W7125972045","https://openalex.org/W7152649144","https://openalex.org/W4391211959","https://openalex.org/W4416529089","https://openalex.org/W4415233581","https://openalex.org/W4407186872","https://openalex.org/W7164992254","https://openalex.org/W7163553499","https://openalex.org/W7154019140"],"abstract_inverted_index":{"Recent":[0],"advances":[1],"in":[2,45],"Chain-of-Thought":[3],"(CoT)":[4],"prompting":[5],"have":[6,18],"substantially":[7],"improved":[8],"the":[9,35,60,73,126,156],"reasoning":[10,65,141,148,172],"capabilities":[11],"of":[12,75,86,182],"large":[13],"language":[14],"models":[15,139],"(LLMs),":[16],"but":[17],"also":[19],"introduced":[20],"their":[21],"computational":[22,92],"efficiency":[23,173],"as":[24],"a":[25,83,112,118,166,179],"new":[26,180],"attack":[27,38,80,110],"surface.":[28],"In":[29],"this":[30,109],"paper,":[31],"we":[32],"propose":[33],"BadThink,":[34],"first":[36],"backdoor":[37],"designed":[39],"to":[40,62,100,124],"deliberately":[41],"induce":[42],"\"overthinking\"":[43],"behavior":[44,127],"CoT-enabled":[46,186],"LLMs":[47],"while":[48,71,97],"ensuring":[49],"stealth.":[50],"When":[51],"activated":[52],"by":[53,128],"carefully":[54],"crafted":[55],"trigger":[56],"prompts,":[57],"BadThink":[58,145],"manipulates":[59],"model":[61],"generate":[63],"inflated":[64],"traces\u2014producing":[66],"unnecessarily":[67],"redundant":[68],"thought":[69],"processes":[70],"preserving":[72],"consistency":[74],"final":[76],"outputs.":[77],"This":[78,163],"subtle":[79],"vector":[81],"creates":[82],"covert":[84],"form":[85],"performance":[87],"degradation":[88],"that":[89,144],"significantly":[90],"increases":[91,147],"costs":[93],"and":[94,140,161],"inference":[95],"time":[96],"remaining":[98,159],"difficult":[99],"detect":[101],"through":[102,111],"conventional":[103],"output":[104],"evaluation":[105],"methods.":[106],"We":[107],"implement":[108],"sophisticated":[113,183],"poisoning-based":[114],"fine-tuning":[115],"strategy,":[116],"employing":[117],"novel":[119],"LLM-based":[120],"iterative":[121],"optimization":[122],"process":[123],"embed":[125],"generating":[129],"highly":[130],"naturalistic":[131],"poisoned":[132],"data.":[133],"Our":[134],"experiments":[135],"on":[136,155],"multiple":[137],"state-of-the-art":[138],"tasks":[142],"show":[143],"consistently":[146],"trace":[149],"lengths\u2014achieving":[150],"an":[151],"over":[152],"17\u00d7":[153],"increase":[154],"MATH-500":[157],"dataset\u2014while":[158],"stealthy":[160],"robust.":[162],"work":[164],"reveals":[165],"critical,":[167],"previously":[168],"unexplored":[169],"vulnerability":[170],"where":[171],"can":[174],"be":[175],"covertly":[176],"manipulated,":[177],"demonstrating":[178],"class":[181],"attacks":[184],"against":[185],"systems.":[187]},"counts_by_year":[],"updated_date":"2026-10-28T10:33:24.094632","created_date":"2026-02-13T00:00:00"}