{"id":"https://openalex.org/W7125972045","doi":"https://doi.org/10.48550/arxiv.2601.19061","title":"Thought-Transfer: Indirect Targeted Poisoning Attacks on Chain-of-Thought Reasoning Models","display_name":"Thought-Transfer: Indirect Targeted Poisoning Attacks on Chain-of-Thought Reasoning Models","publication_year":2026,"publication_date":"2026-01-27","ids":{"openalex":"https://openalex.org/W7125972045","doi":"https://doi.org/10.48550/arxiv.2601.19061"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2601.19061","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2601.19061","pdf_url":"https://arxiv.org/pdf/2601.19061","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2601.19061","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5045507972","display_name":"Harsh Chaudhari","orcid":"https://orcid.org/0009-0002-0430-2025","observed_orcids":["https://orcid.org/0009-0002-0430-2025"]},"institutions":[{"id":"https://openalex.org/I12912129","display_name":"Northeastern University","ror":"https://ror.org/04t5xt781","country_code":"US","type":"education","lineage":["https://openalex.org/I12912129"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Chaudhari, Harsh","raw_affiliation_strings":["Northeastern University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Northeastern University","institution_ids":["https://openalex.org/I12912129"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5015757953","display_name":"Ethan Rathbun","orcid":"https://orcid.org/0000-0002-5437-2489","observed_orcids":["https://orcid.org/0000-0002-5437-2489"]},"institutions":[{"id":"https://openalex.org/I12912129","display_name":"Northeastern University","ror":"https://ror.org/04t5xt781","country_code":"US","type":"education","lineage":["https://openalex.org/I12912129"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Rathbun, Ethan","raw_affiliation_strings":["Northeastern University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Northeastern University","institution_ids":["https://openalex.org/I12912129"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5099173797","display_name":"Hanna Foerster","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I241749","display_name":"University of Cambridge","ror":"https://ror.org/013meh722","country_code":"GB","type":"education","lineage":["https://openalex.org/I241749"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Foerster, Hanna","raw_affiliation_strings":["University of Cambridge"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Cambridge","institution_ids":["https://openalex.org/I241749"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5020715720","display_name":"Jamie Hayes","orcid":"https://orcid.org/0009-0008-5460-6290","observed_orcids":["https://orcid.org/0009-0008-5460-6290"]},"institutions":[{"id":"https://openalex.org/I4210090411","display_name":"Google DeepMind (United Kingdom)","ror":"https://ror.org/00971b260","country_code":"GB","type":"company","lineage":["https://openalex.org/I4210090411","https://openalex.org/I4210128969"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Hayes, Jamie","raw_affiliation_strings":["Google DeepMind"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Google DeepMind","institution_ids":["https://openalex.org/I4210090411"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5054655342","display_name":"Matthew Jagielski","orcid":"https://orcid.org/0000-0002-9749-0696","observed_orcids":["https://orcid.org/0000-0002-9749-0696"]},"institutions":[{"id":"https://openalex.org/I4387930290","display_name":"Anthropic (United States)","ror":"https://ror.org/056y0v115","country_code":"US","type":"company","lineage":["https://openalex.org/I4387930290"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Jagielski, Matthew","raw_affiliation_strings":["Anthropic"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Anthropic","institution_ids":["https://openalex.org/I4387930290"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5059480732","display_name":"Milad Nasr","orcid":"https://orcid.org/0000-0002-1913-6157","observed_orcids":["https://orcid.org/0000-0002-1913-6157"]},"institutions":[{"id":"https://openalex.org/I4210161460","display_name":"OpenAI (United States)","ror":"https://ror.org/05wx9n238","country_code":"US","type":"company","lineage":["https://openalex.org/I4210161460"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Nasr, Milad","raw_affiliation_strings":["OpenAI"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"OpenAI","institution_ids":["https://openalex.org/I4210161460"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5069844959","display_name":"Ilia Shumailov","orcid":"https://orcid.org/0000-0003-3100-0727","observed_orcids":["https://orcid.org/0000-0003-3100-0727"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shumailov, Ilia","raw_affiliation_strings":["AI Sequrity"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"AI Sequrity","institution_ids":[]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5035574749","display_name":"Alina Oprea","orcid":"https://orcid.org/0000-0002-4979-5292","observed_orcids":["https://orcid.org/0000-0002-4979-5292"]},"institutions":[{"id":"https://openalex.org/I12912129","display_name":"Northeastern University","ror":"https://ror.org/04t5xt781","country_code":"US","type":"education","lineage":["https://openalex.org/I12912129"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Oprea, Alina","raw_affiliation_strings":["Northeastern University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Northeastern University","institution_ids":["https://openalex.org/I12912129"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":5,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":5,"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/T10764","display_name":"Privacy-Preserving Technologies in Data","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-reasoning","display_name":"chain-of-thought reasoning","score":0.8989999890327454},{"id":"https://openalex.org/keywords/adversarial-machine-learning","display_name":"adversarial learning","score":0.6539999842643738},{"id":"https://openalex.org/keywords/large-language-models","display_name":"large language models","score":0.902999997138977},{"id":"https://openalex.org/keywords/backdoor-attacks","display_name":"backdoor attacks","score":0.7620000243186951},{"id":"https://openalex.org/keywords/reasoning-traces","display_name":"reasoning traces","score":0.6150000095367432},{"id":"https://openalex.org/keywords/fine-tuning","display_name":"fine-tuning","score":0.42100000381469727}],"concepts":[{"id":"https://openalex.org/C2781045450","wikidata":"https://www.wikidata.org/wiki/Q254569","display_name":"Backdoor","level":2,"score":0.7551000118255615},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.7361000180244446},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6543999910354614},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.5396999716758728},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.460099995136261},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.45989999175071716},{"id":"https://openalex.org/C2780289543","wikidata":"https://www.wikidata.org/wiki/Q424630","display_name":"Accident (philosophy)","level":2,"score":0.45329999923706055},{"id":"https://openalex.org/C20162079","wikidata":"https://www.wikidata.org/wiki/Q1151406","display_name":"Case-based reasoning","level":2,"score":0.43389999866485596},{"id":"https://openalex.org/C2777212361","wikidata":"https://www.wikidata.org/wiki/Q5127848","display_name":"Class (philosophy)","level":2,"score":0.43059998750686646},{"id":"https://openalex.org/C115086926","wikidata":"https://www.wikidata.org/wiki/Q17004651","display_name":"Causal reasoning","level":3,"score":0.4162999987602234},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.3693999946117401},{"id":"https://openalex.org/C107457646","wikidata":"https://www.wikidata.org/wiki/Q207434","display_name":"Human\u2013computer interaction","level":1,"score":0.33180001378059387},{"id":"https://openalex.org/C159032336","wikidata":"https://www.wikidata.org/wiki/Q2488768","display_name":"Non-monotonic logic","level":2,"score":0.3224000036716461},{"id":"https://openalex.org/C89288958","wikidata":"https://www.wikidata.org/wiki/Q7301504","display_name":"Reasoning system","level":2,"score":0.3197999894618988},{"id":"https://openalex.org/C18762648","wikidata":"https://www.wikidata.org/wiki/Q42213","display_name":"Work (physics)","level":2,"score":0.3140999972820282},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3125999867916107},{"id":"https://openalex.org/C97364631","wikidata":"https://www.wikidata.org/wiki/Q484284","display_name":"Deductive reasoning","level":2,"score":0.30399999022483826},{"id":"https://openalex.org/C29122968","wikidata":"https://www.wikidata.org/wiki/Q1414816","display_name":"Incentive","level":2,"score":0.2919999957084656},{"id":"https://openalex.org/C109359841","wikidata":"https://www.wikidata.org/wiki/Q728944","display_name":"Inclusion (mineral)","level":2,"score":0.2912999987602234},{"id":"https://openalex.org/C60643870","wikidata":"https://www.wikidata.org/wiki/Q1949683","display_name":"Deterrence theory","level":2,"score":0.2888999879360199},{"id":"https://openalex.org/C86827895","wikidata":"https://www.wikidata.org/wiki/Q7098582","display_name":"Opportunistic reasoning","level":4,"score":0.28110000491142273},{"id":"https://openalex.org/C175154964","wikidata":"https://www.wikidata.org/wiki/Q380077","display_name":"Task analysis","level":3,"score":0.27970001101493835},{"id":"https://openalex.org/C3017944768","wikidata":"https://www.wikidata.org/wiki/Q1450463","display_name":"Poison control","level":2,"score":0.2709999978542328},{"id":"https://openalex.org/C2776325391","wikidata":"https://www.wikidata.org/wiki/Q6917865","display_name":"Motivated reasoning","level":3,"score":0.2619999945163727},{"id":"https://openalex.org/C37335422","wikidata":"https://www.wikidata.org/wiki/Q6888134","display_name":"Model-based reasoning","level":3,"score":0.25529998540878296},{"id":"https://openalex.org/C195344581","wikidata":"https://www.wikidata.org/wiki/Q2555318","display_name":"Automated reasoning","level":2,"score":0.2547999918460846}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2601.19061","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2601.19061","pdf_url":"https://arxiv.org/pdf/2601.19061","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2601.19061","doi":"https://doi.org/10.48550/arxiv.2601.19061","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2601.19061","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2601.19061","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2601.19061","pdf_url":"https://arxiv.org/pdf/2601.19061","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[{"id":"https://openalex.org/G279616255","display_name":"NeTS: Medium: Resilient-by-Design Data-Driven NextG Open Radio Access Networks","funder_award_id":"2312875","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G5259331294","display_name":null,"funder_award_id":"W911NF","funder_id":"https://openalex.org/F4320338295","funder_display_name":"Army Research Laboratory"},{"id":"https://openalex.org/G7731344590","display_name":"SaTC: TTP: Small: Poisoning-Resilient Machine Learning Models for Threat Detection","funder_award_id":"2331081","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"},{"id":"https://openalex.org/F4320315784","display_name":"U.S. Army Combat Capabilities Development Command Soldier Center","ror":"https://ror.org/02rdkx920"},{"id":"https://openalex.org/F4320337807","display_name":"U.S. Army Combat Capabilities Development Command","ror":null},{"id":"https://openalex.org/F4320338295","display_name":"Army Research Laboratory","ror":"https://ror.org/011hc8f90"},{"id":"https://openalex.org/F4320338456","display_name":"DEVCOM Army Research Laboratory","ror":null}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7125972045.pdf","grobid_xml":"https://content.openalex.org/works/W7125972045.grobid-xml"},"referenced_works_count":12,"referenced_works":["https://openalex.org/W2748789698","https://openalex.org/W3210904070","https://openalex.org/W4386270999","https://openalex.org/W4405899099","https://openalex.org/W4407092861","https://openalex.org/W4407690019","https://openalex.org/W4407759168","https://openalex.org/W4414755425","https://openalex.org/W4415067413","https://openalex.org/W4415108257","https://openalex.org/W4415233581","https://openalex.org/W4416529089"],"related_works":["https://openalex.org/W4416339359","https://openalex.org/W7164992254","https://openalex.org/W7169765668","https://openalex.org/W7152649144","https://openalex.org/W7163553499","https://openalex.org/W4414968312","https://openalex.org/W4416529089","https://openalex.org/W4414755425","https://openalex.org/W4402683743","https://openalex.org/W7172513225"],"abstract_inverted_index":{"Chain-of-Thought":[0],"(CoT)":[1],"reasoning":[2,18,30,51,78,100,193,214,225],"has":[3],"emerged":[4],"as":[5],"a":[6,91,106,115,127,147,208,219],"powerful":[7],"technique":[8],"for":[9,20,26,207],"enhancing":[10],"large":[11],"language":[12],"models'":[13],"capabilities":[14],"by":[15,109,130,200,224,232],"generating":[16],"intermediate":[17],"steps":[19],"complex":[21],"tasks.":[22],"A":[23],"common":[24],"practice":[25],"equipping":[27],"LLMs":[28],"with":[29,76],"is":[31,228],"to":[32,86,210],"fine-tune":[33],"pre-trained":[34],"models":[35,101],"using":[36],"CoT":[37,111,136],"datasets":[38],"from":[39,114],"public":[40],"repositories":[41],"like":[42],"HuggingFace,":[43],"which":[44,227],"creates":[45],"new":[46,92],"attack":[47,120],"vectors":[48],"targeting":[49],"the":[50,59,83,123,133,140,165,197],"traces":[52,112],"themselves.":[53],"While":[54],"prior":[55,154],"works":[56],"have":[57],"shown":[58],"possibility":[60],"of":[61,73,94,105,149],"mounting":[62],"backdoor":[63],"attacks":[64,69,98,157],"in":[65,82,99,146,164,176,188],"CoT-based":[66],"models,":[67,226],"these":[68],"require":[70,160],"explicit":[71],"inclusion":[72],"triggered":[74],"queries":[75,141],"flawed":[77],"and":[79,142],"incorrect":[80],"answers":[81,143],"training":[84,134],"set":[85],"succeed.":[87],"Our":[88,118,216],"work":[89],"unveils":[90],"class":[93],"Indirect":[95],"Targeted":[96],"Poisoning":[97],"that":[102,158,170,184],"manipulate":[103],"responses":[104],"target":[107,128,161],"task":[108,129,162],"transferring":[110],"learned":[113],"different":[116,182],"task.":[117],"\"Thought-Transfer\"":[119],"can":[121],"influence":[122],"LLM":[124],"output":[125],"on":[126,191,202],"manipulating":[131],"only":[132],"samples'":[135],"traces,":[137],"while":[138],"leaving":[139],"unchanged,":[144],"resulting":[145],"form":[148],"``clean":[150],"label''":[151],"poisoning.":[152],"Unlike":[153],"targeted":[155,178],"poisoning":[156],"explicitly":[159],"samples":[163],"poisoned":[166,192,213],"data,":[167],"we":[168],"demonstrate":[169],"thought-transfer":[171],"achieves":[172],"70%":[173],"success":[174],"rates":[175],"injecting":[177],"behaviors":[179],"into":[180],"entirely":[181],"domains":[183],"are":[185],"never":[186],"present":[187],"training.":[189],"Training":[190],"data":[194],"also":[195],"improves":[196],"model's":[198],"performance":[199],"10-15%":[201],"multiple":[203],"benchmarks,":[204],"providing":[205],"incentives":[206],"user":[209],"use":[211],"our":[212],"dataset.":[215],"findings":[217],"reveal":[218],"novel":[220],"threat":[221],"vector":[222],"enabled":[223],"not":[229],"easily":[230],"defended":[231],"existing":[233],"mitigations.":[234]},"counts_by_year":[{"year":2026,"cited_by_count":5}],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2026-01-29T00:00:00"}