{"id":"https://openalex.org/W7225388602","doi":"https://doi.org/10.48550/arxiv.2611.08438","title":"Understanding AI from the Inside: The State of Mechanistic Interpretability for AI Safety","display_name":"Understanding AI from the Inside: The State of Mechanistic Interpretability for AI Safety","publication_year":2026,"publication_date":"2026-11-05","ids":{"openalex":"https://openalex.org/W7225388602","doi":"https://doi.org/10.48550/arxiv.2611.08438"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2611.08438","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2611.08438","pdf_url":"https://arxiv.org/pdf/2611.08438","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/2611.08438","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5050436152","display_name":"Fazl Barez","orcid":"https://orcid.org/0009-0008-1889-6577","observed_orcids":["https://orcid.org/0009-0008-1889-6577"]},"institutions":[{"id":"https://openalex.org/I40120149","display_name":"University of Oxford","ror":"https://ror.org/052gg0110","country_code":"GB","type":"education","lineage":["https://openalex.org/I40120149"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Barez, Fazl","raw_affiliation_strings":["University of Oxford"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Oxford","institution_ids":["https://openalex.org/I40120149"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5161047396","display_name":"Tung-Yu Wu","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I40120149","display_name":"University of Oxford","ror":"https://ror.org/052gg0110","country_code":"GB","type":"education","lineage":["https://openalex.org/I40120149"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Wu, Tung-Yu","raw_affiliation_strings":["University of Oxford"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Oxford","institution_ids":["https://openalex.org/I40120149"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5048081395","display_name":"Hunar Batra","orcid":"https://orcid.org/0000-0001-8818-5959","observed_orcids":["https://orcid.org/0000-0001-8818-5959"]},"institutions":[{"id":"https://openalex.org/I40120149","display_name":"University of Oxford","ror":"https://ror.org/052gg0110","country_code":"GB","type":"education","lineage":["https://openalex.org/I40120149"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Batra, Hunar","raw_affiliation_strings":["University of Oxford"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Oxford","institution_ids":["https://openalex.org/I40120149"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5104314485","display_name":"Gabriele Dominici","orcid":"https://orcid.org/0009-0009-1955-0778","observed_orcids":["https://orcid.org/0009-0009-1955-0778"]},"institutions":[{"id":"https://openalex.org/I57201433","display_name":"Universit\u00e0 della Svizzera italiana","ror":"https://ror.org/03c4atk17","country_code":"CH","type":"education","lineage":["https://openalex.org/I57201433"]}],"countries":["CH"],"is_corresponding":false,"raw_author_name":"Dominici, Gabriele","raw_affiliation_strings":["Universit\u00e0 della Svizzera italiana"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Universit\u00e0 della Svizzera italiana","institution_ids":["https://openalex.org/I57201433"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5120409390","display_name":"Florent Draye","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I4210135521","display_name":"Max Planck Institute for Intelligent Systems","ror":"https://ror.org/04fq9j139","country_code":"DE","type":"facility","lineage":["https://openalex.org/I149899117","https://openalex.org/I4210135521"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Draye, Florent","raw_affiliation_strings":["Max Planck Institute for Intelligent Systems"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Max Planck Institute for Intelligent Systems","institution_ids":["https://openalex.org/I4210135521"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5081462507","display_name":"Yanai Elazar","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I13955877","display_name":"Bar-Ilan University","ror":"https://ror.org/03kgsv495","country_code":"IL","type":"education","lineage":["https://openalex.org/I13955877"]}],"countries":["IL"],"is_corresponding":false,"raw_author_name":"Elazar, Yanai","raw_affiliation_strings":["Bar-Ilan University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Bar-Ilan University","institution_ids":["https://openalex.org/I13955877"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134098489","display_name":"Damiano Fornasiere","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I129902397","display_name":"Dalhousie University","ror":"https://ror.org/01e6qks80","country_code":"CA","type":"education","lineage":["https://openalex.org/I129902397"]}],"countries":["CA"],"is_corresponding":false,"raw_author_name":"Fornasiere, Damiano","raw_affiliation_strings":["Dalhousie University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Dalhousie University","institution_ids":["https://openalex.org/I129902397"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5159850686","display_name":"Ant\u00eda Garc\u00eda","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I129902397","display_name":"Dalhousie University","ror":"https://ror.org/01e6qks80","country_code":"CA","type":"education","lineage":["https://openalex.org/I129902397"]}],"countries":["CA"],"is_corresponding":false,"raw_author_name":"Garc\u00eda, Ant\u00eda","raw_affiliation_strings":["Dalhousie University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Dalhousie University","institution_ids":["https://openalex.org/I129902397"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137832082","display_name":"Abir Harrasse","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I129902397","display_name":"Dalhousie University","ror":"https://ror.org/01e6qks80","country_code":"CA","type":"education","lineage":["https://openalex.org/I129902397"]}],"countries":["CA"],"is_corresponding":false,"raw_author_name":"Harrasse, Abir","raw_affiliation_strings":["Dalhousie University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Dalhousie University","institution_ids":["https://openalex.org/I129902397"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5161068898","display_name":"Riley Harris","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I40120149","display_name":"University of Oxford","ror":"https://ror.org/052gg0110","country_code":"GB","type":"education","lineage":["https://openalex.org/I40120149"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Harris, Riley","raw_affiliation_strings":["University of Oxford"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Oxford","institution_ids":["https://openalex.org/I40120149"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5078850040","display_name":"Sara Hooker","orcid":"https://orcid.org/0000-0002-0190-6459","observed_orcids":["https://orcid.org/0000-0002-0190-6459"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hooker, Sara","raw_affiliation_strings":["Adaption Labs"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Adaption Labs","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5058869060","display_name":"Michael Lan","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I129902397","display_name":"Dalhousie University","ror":"https://ror.org/01e6qks80","country_code":"CA","type":"education","lineage":["https://openalex.org/I129902397"]}],"countries":["CA"],"is_corresponding":false,"raw_author_name":"Lan, Michael","raw_affiliation_strings":["Dalhousie University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Dalhousie University","institution_ids":["https://openalex.org/I129902397"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5161047569","display_name":"Belinda Z. Li","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I63966007","display_name":"Massachusetts Institute of Technology","ror":"https://ror.org/042nb2s44","country_code":"US","type":"education","lineage":["https://openalex.org/I63966007"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Li, Belinda Z.","raw_affiliation_strings":["MIT"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"MIT","institution_ids":["https://openalex.org/I63966007"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5120089970","display_name":"Aaron Maiwald","orcid":"https://orcid.org/0009-0000-2897-2862","observed_orcids":["https://orcid.org/0009-0000-2897-2862"]},"institutions":[{"id":"https://openalex.org/I40120149","display_name":"University of Oxford","ror":"https://ror.org/052gg0110","country_code":"GB","type":"education","lineage":["https://openalex.org/I40120149"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Maiwald, Aaron","raw_affiliation_strings":["University of Oxford"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Oxford","institution_ids":["https://openalex.org/I40120149"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5119862711","display_name":"Narmeen Fatimah Oozeer","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I129902397","display_name":"Dalhousie University","ror":"https://ror.org/01e6qks80","country_code":"CA","type":"education","lineage":["https://openalex.org/I129902397"]}],"countries":["CA"],"is_corresponding":false,"raw_author_name":"Oozeer, Narmeen Fatimah","raw_affiliation_strings":["Dalhousie University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Dalhousie University","institution_ids":["https://openalex.org/I129902397"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5010251681","display_name":"Naomi Saphra","orcid":"https://orcid.org/0000-0001-8607-3515","observed_orcids":["https://orcid.org/0000-0001-8607-3515"]},"institutions":[{"id":"https://openalex.org/I111088046","display_name":"Boston University","ror":"https://ror.org/05qwgg493","country_code":"US","type":"education","lineage":["https://openalex.org/I111088046"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Saphra, Naomi","raw_affiliation_strings":["Boston University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Boston University","institution_ids":["https://openalex.org/I111088046"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5099987828","display_name":"Or David Shafran","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shafran, Or David","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5056036228","display_name":"Adi Simhi","orcid":"https://orcid.org/0000-0003-1730-7087","observed_orcids":["https://orcid.org/0000-0003-1730-7087"]},"institutions":[{"id":"https://openalex.org/I162714631","display_name":"George Mason University","ror":"https://ror.org/02jqj7156","country_code":"US","type":"education","lineage":["https://openalex.org/I162714631"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Simhi, Adi","raw_affiliation_strings":["Technion 13 George Mason University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Technion 13 George Mason University","institution_ids":["https://openalex.org/I162714631"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5053047897","display_name":"Robert Trager","orcid":"https://orcid.org/0000-0002-6201-2811","observed_orcids":["https://orcid.org/0000-0002-6201-2811"]},"institutions":[{"id":"https://openalex.org/I40120149","display_name":"University of Oxford","ror":"https://ror.org/052gg0110","country_code":"GB","type":"education","lineage":["https://openalex.org/I40120149"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Trager, Robert","raw_affiliation_strings":["University of Oxford"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Oxford","institution_ids":["https://openalex.org/I40120149"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5094196399","display_name":"Shriyash Upadhyay","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I129902397","display_name":"Dalhousie University","ror":"https://ror.org/01e6qks80","country_code":"CA","type":"education","lineage":["https://openalex.org/I129902397"]}],"countries":["CA"],"is_corresponding":false,"raw_author_name":"Upadhyay, Shriyash Kaustubh","raw_affiliation_strings":["Dalhousie University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Dalhousie University","institution_ids":["https://openalex.org/I129902397"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5161078064","display_name":"Ziyu Yao","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I129902397","display_name":"Dalhousie University","ror":"https://ror.org/01e6qks80","country_code":"CA","type":"education","lineage":["https://openalex.org/I129902397"]}],"countries":["CA"],"is_corresponding":false,"raw_author_name":"Yao, Ziyu","raw_affiliation_strings":["Dalhousie University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Dalhousie University","institution_ids":["https://openalex.org/I129902397"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5161057669","display_name":"Yu Zhao","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I40120149","display_name":"University of Oxford","ror":"https://ror.org/052gg0110","country_code":"GB","type":"education","lineage":["https://openalex.org/I40120149"]},{"id":"https://openalex.org/I98677209","display_name":"University of Edinburgh","ror":"https://ror.org/01nrxwf90","country_code":"GB","type":"education","lineage":["https://openalex.org/I98677209"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Zhao, Yu","raw_affiliation_strings":["University of Edinburgh","University of Oxford"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Edinburgh","institution_ids":["https://openalex.org/I98677209"]},{"raw_affiliation_string":"University of Oxford","institution_ids":["https://openalex.org/I40120149"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5030503109","display_name":"Shay B. Cohen","orcid":"https://orcid.org/0000-0003-4753-8353","observed_orcids":["https://orcid.org/0000-0003-4753-8353"]},"institutions":[{"id":"https://openalex.org/I98677209","display_name":"University of Edinburgh","ror":"https://ror.org/01nrxwf90","country_code":"GB","type":"education","lineage":["https://openalex.org/I98677209"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Cohen, Shay B.","raw_affiliation_strings":["University of Edinburgh"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Edinburgh","institution_ids":["https://openalex.org/I98677209"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5065717258","display_name":"Mor Geva","orcid":"https://orcid.org/0000-0001-9529-6315","observed_orcids":["https://orcid.org/0000-0001-9529-6315"]},"institutions":[{"id":"https://openalex.org/I16391192","display_name":"Tel Aviv University","ror":"https://ror.org/04mhzgx49","country_code":"IL","type":"education","lineage":["https://openalex.org/I16391192"]}],"countries":["IL"],"is_corresponding":false,"raw_author_name":"Geva, Mor","raw_affiliation_strings":["Tel Aviv University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Tel Aviv University","institution_ids":["https://openalex.org/I16391192"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5086198262","display_name":"Yoshua Bengio","orcid":"https://orcid.org/0000-0002-9322-3515","observed_orcids":["https://orcid.org/0000-0002-9322-3515"]},"institutions":[{"id":"https://openalex.org/I70931966","display_name":"Universit\u00e9 de Montr\u00e9al","ror":"https://ror.org/0161xgx34","country_code":"CA","type":"education","lineage":["https://openalex.org/I70931966"]}],"countries":["CA"],"is_corresponding":false,"raw_author_name":"Bengio, Yoshua","raw_affiliation_strings":["Mila , Universit\u00e9 de Montr\u00e9al"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Mila , Universit\u00e9 de Montr\u00e9al","institution_ids":["https://openalex.org/I70931966"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5076733732","display_name":"Stella Biderman","orcid":"https://orcid.org/0000-0001-8228-1042","observed_orcids":["https://orcid.org/0000-0001-8228-1042"]},"institutions":[{"id":"https://openalex.org/I120250893","display_name":"New York University Abu Dhabi","ror":"https://ror.org/00e5k0821","country_code":"AE","type":"education","lineage":["https://openalex.org/I120250893","https://openalex.org/I57206974"]}],"countries":["AE"],"is_corresponding":false,"raw_author_name":"Biderman, Stella","raw_affiliation_strings":["EleutherAI 17 New York University Abu Dhabi &"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"EleutherAI 17 New York University Abu Dhabi &","institution_ids":["https://openalex.org/I120250893"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5042954793","display_name":"Hassan Sajjad","orcid":"https://orcid.org/0000-0002-8584-6595","observed_orcids":["https://orcid.org/0000-0002-8584-6595"]},"institutions":[{"id":"https://openalex.org/I129902397","display_name":"Dalhousie University","ror":"https://ror.org/01e6qks80","country_code":"CA","type":"education","lineage":["https://openalex.org/I129902397"]}],"countries":["CA"],"is_corresponding":false,"raw_author_name":"Sajjad, Hassan","raw_affiliation_strings":["Dalhousie University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Dalhousie University","institution_ids":["https://openalex.org/I129902397"]}]}],"institutions":[],"countries_distinct_count":7,"institutions_distinct_count":12,"corresponding_author_ids":[],"corresponding_institution_ids":[],"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":false,"primary_topic":{"id":"https://openalex.org/T12026","display_name":"Explainable Artificial Intelligence (XAI)","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/T12026","display_name":"Explainable Artificial Intelligence (XAI)","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/T11689","display_name":"Adversarial Robustness in Machine Learning","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/T10883","display_name":"Ethics and Social Impacts of AI","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"}}],"keywords":[{"id":"https://openalex.org/keywords/mechanistic-interpretability","display_name":"mechanistic interpretability","score":0.9990000128746033},{"id":"https://openalex.org/keywords/ai-safety","display_name":"AI safety","score":0.996999979019165},{"id":"https://openalex.org/keywords/neural-network","display_name":"artificial neural networks","score":0.781000018119812},{"id":"https://openalex.org/keywords/internal-representations","display_name":"internal representations","score":0.6179999709129333},{"id":"https://openalex.org/keywords/transformer-circuits","display_name":"transformer circuits","score":0.550000011920929},{"id":"https://openalex.org/keywords/feature-extraction","display_name":"feature extraction","score":0.6880000233650208},{"id":"https://openalex.org/keywords/polysemanticity","display_name":"polysemanticity","score":0.9210000038146973},{"id":"https://openalex.org/keywords/training-data-attribution","display_name":"training data attribution","score":0.6639999747276306},{"id":"https://openalex.org/keywords/learning-dynamics","display_name":"learning dynamics","score":0.753000020980835},{"id":"https://openalex.org/keywords/knowledge-editing","display_name":"knowledge editing","score":0.7609999775886536},{"id":"https://openalex.org/keywords/steering","display_name":"steering","score":0.8240000009536743},{"id":"https://openalex.org/keywords/monitoring","display_name":"monitoring","score":0.7630000114440918},{"id":"https://openalex.org/keywords/safety-behavior","display_name":"safety behavior","score":0.5370000004768372}],"concepts":[{"id":"https://openalex.org/C2781067378","wikidata":"https://www.wikidata.org/wiki/Q17027399","display_name":"Interpretability","level":2,"score":0.9739000201225281},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6319000124931335},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5260000228881836},{"id":"https://openalex.org/C2522767166","wikidata":"https://www.wikidata.org/wiki/Q2374463","display_name":"Data science","level":1,"score":0.4595000147819519},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.45419999957084656},{"id":"https://openalex.org/C2777363581","wikidata":"https://www.wikidata.org/wiki/Q15098235","display_name":"Harm","level":2,"score":0.42410001158714294},{"id":"https://openalex.org/C188147891","wikidata":"https://www.wikidata.org/wiki/Q147638","display_name":"Cognitive science","level":1,"score":0.39660000801086426},{"id":"https://openalex.org/C132010649","wikidata":"https://www.wikidata.org/wiki/Q189222","display_name":"Intuition","level":2,"score":0.37950000166893005},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3596000075340271},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.326200008392334},{"id":"https://openalex.org/C89611455","wikidata":"https://www.wikidata.org/wiki/Q6804646","display_name":"Mechanism (biology)","level":2,"score":0.2955000102519989},{"id":"https://openalex.org/C539667460","wikidata":"https://www.wikidata.org/wiki/Q2414942","display_name":"Management science","level":1,"score":0.2856999933719635},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.25940001010894775},{"id":"https://openalex.org/C112930515","wikidata":"https://www.wikidata.org/wiki/Q4389547","display_name":"Risk analysis (engineering)","level":1,"score":0.25929999351501465},{"id":"https://openalex.org/C47822265","wikidata":"https://www.wikidata.org/wiki/Q854457","display_name":"Complex system","level":2,"score":0.25360000133514404}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2611.08438","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2611.08438","pdf_url":"https://arxiv.org/pdf/2611.08438","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.2611.08438","doi":"https://doi.org/10.48550/arxiv.2611.08438","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2611.08438","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:2611.08438","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2611.08438","pdf_url":"https://arxiv.org/pdf/2611.08438","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":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7225388602.pdf","grobid_xml":"https://content.openalex.org/works/W7225388602.grobid-xml"},"referenced_works_count":195,"referenced_works":["https://openalex.org/W1787224781","https://openalex.org/W1902027874","https://openalex.org/W1951216520","https://openalex.org/W1982370770","https://openalex.org/W1986931325","https://openalex.org/W1994616650","https://openalex.org/W2053987251","https://openalex.org/W2071128523","https://openalex.org/W2085927826","https://openalex.org/W2087946919","https://openalex.org/W2105464873","https://openalex.org/W2140499889","https://openalex.org/W2145889472","https://openalex.org/W2151693816","https://openalex.org/W2168103112","https://openalex.org/W2242818861","https://openalex.org/W2250382531","https://openalex.org/W2264869955","https://openalex.org/W2273396394","https://openalex.org/W2516809705","https://openalex.org/W2549835527","https://openalex.org/W2563574619","https://openalex.org/W2605717780","https://openalex.org/W2626639386","https://openalex.org/W2775461895","https://openalex.org/W2888169584","https://openalex.org/W2888922637","https://openalex.org/W2934842096","https://openalex.org/W2946417913","https://openalex.org/W2946794439","https://openalex.org/W2949202705","https://openalex.org/W2949603537","https://openalex.org/W2951286828","https://openalex.org/W2953337107","https://openalex.org/W2963341956","https://openalex.org/W2963400886","https://openalex.org/W2964159778","https://openalex.org/W2964204621","https://openalex.org/W2968210605","https://openalex.org/W2970862333","https://openalex.org/W3001279689","https://openalex.org/W3006439205","https://openalex.org/W3006881356","https://openalex.org/W3010694149","https://openalex.org/W3013891696","https://openalex.org/W3015575765","https://openalex.org/W3016970897","https://openalex.org/W3037932933","https://openalex.org/W3082665562","https://openalex.org/W3117576675","https://openalex.org/W3118781290","https://openalex.org/W3152884768","https://openalex.org/W3159900299","https://openalex.org/W3166035876","https://openalex.org/W3170470779","https://openalex.org/W3177813494","https://openalex.org/W3196986263","https://openalex.org/W3210904070","https://openalex.org/W4223908421","https://openalex.org/W4226434736","https://openalex.org/W4226462869","https://openalex.org/W4229494842","https://openalex.org/W4288804650","https://openalex.org/W4296413526","https://openalex.org/W4308885736","https://openalex.org/W4311991106","https://openalex.org/W4322616316","https://openalex.org/W4322718191","https://openalex.org/W4327526719","https://openalex.org/W4365597196","https://openalex.org/W4378976798","https://openalex.org/W4381245716","https://openalex.org/W4385473442","https://openalex.org/W4385569978","https://openalex.org/W4385570455","https://openalex.org/W4385571289","https://openalex.org/W4385571791","https://openalex.org/W4385572928","https://openalex.org/W4385681731","https://openalex.org/W4385894687","https://openalex.org/W4386436506","https://openalex.org/W4386559757","https://openalex.org/W4389518760","https://openalex.org/W4389524555","https://openalex.org/W4389617257","https://openalex.org/W4389708672","https://openalex.org/W4389713766","https://openalex.org/W4390043316","https://openalex.org/W4390490761","https://openalex.org/W4390723197","https://openalex.org/W4390833079","https://openalex.org/W4391334942","https://openalex.org/W4391591214","https://openalex.org/W4396648991","https://openalex.org/W4399510412","https://openalex.org/W4399795554","https://openalex.org/W4400065290","https://openalex.org/W4400374124","https://openalex.org/W4400375635","https://openalex.org/W4401024111","https://openalex.org/W4401042689","https://openalex.org/W4401200976","https://openalex.org/W4402669924","https://openalex.org/W4402671301","https://openalex.org/W4402671334","https://openalex.org/W4402671619","https://openalex.org/W4402671766","https://openalex.org/W4402671828","https://openalex.org/W4402671950","https://openalex.org/W4403345627","https://openalex.org/W4403593439","https://openalex.org/W4404314166","https://openalex.org/W4404354156","https://openalex.org/W4404752287","https://openalex.org/W4404782389","https://openalex.org/W4404783016","https://openalex.org/W4404783416","https://openalex.org/W4404783717","https://openalex.org/W4404784049","https://openalex.org/W4404792961","https://openalex.org/W4404987046","https://openalex.org/W4405627013","https://openalex.org/W4405768303","https://openalex.org/W4405903187","https://openalex.org/W4406071632","https://openalex.org/W4406840771","https://openalex.org/W4406879729","https://openalex.org/W4407093342","https://openalex.org/W4407093355","https://openalex.org/W4408868638","https://openalex.org/W4409361279","https://openalex.org/W4412886783","https://openalex.org/W4412886802","https://openalex.org/W4412887974","https://openalex.org/W4412888620","https://openalex.org/W4412889953","https://openalex.org/W4412945693","https://openalex.org/W4414586248","https://openalex.org/W4414771228","https://openalex.org/W4414844329","https://openalex.org/W4414939355","https://openalex.org/W4414946584","https://openalex.org/W4415063744","https://openalex.org/W4415181430","https://openalex.org/W4415244716","https://openalex.org/W4416034000","https://openalex.org/W4416034653","https://openalex.org/W4416034678","https://openalex.org/W4416035347","https://openalex.org/W4416035715","https://openalex.org/W4416036025","https://openalex.org/W4416036098","https://openalex.org/W4416091752","https://openalex.org/W4416098779","https://openalex.org/W4416160875","https://openalex.org/W4416185867","https://openalex.org/W4416191647","https://openalex.org/W4416329855","https://openalex.org/W4417114549","https://openalex.org/W4417143258","https://openalex.org/W4417144098","https://openalex.org/W4417300717","https://openalex.org/W4417357092","https://openalex.org/W4417448492","https://openalex.org/W6910566006","https://openalex.org/W6948017749","https://openalex.org/W7106695004","https://openalex.org/W7114786565","https://openalex.org/W7119916356","https://openalex.org/W7126156023","https://openalex.org/W7126375801","https://openalex.org/W7127179963","https://openalex.org/W7128000853","https://openalex.org/W7128383720","https://openalex.org/W7128716986","https://openalex.org/W7129336288","https://openalex.org/W7131412137","https://openalex.org/W7133219754","https://openalex.org/W7133501764","https://openalex.org/W7154461809","https://openalex.org/W7155039456","https://openalex.org/W7155052370","https://openalex.org/W7155554729","https://openalex.org/W7160261982","https://openalex.org/W7160544910","https://openalex.org/W7164362211","https://openalex.org/W7165735320","https://openalex.org/W7166653112","https://openalex.org/W7166813420","https://openalex.org/W7166842394","https://openalex.org/W7166846670","https://openalex.org/W7166886272","https://openalex.org/W7202206632","https://openalex.org/W7203714722","https://openalex.org/W7220387322"],"related_works":["https://openalex.org/W4395065622","https://openalex.org/W7223013230","https://openalex.org/W7125584057","https://openalex.org/W7166811405","https://openalex.org/W7108329787","https://openalex.org/W7169754304","https://openalex.org/W4403315440","https://openalex.org/W7118642194","https://openalex.org/W7167836443","https://openalex.org/W7119998135"],"abstract_inverted_index":{"Neural":[0],"networks":[1],"are":[2],"likely":[3],"to":[4,24,44,64,102,223,247],"be":[5],"among":[6],"the":[7,35,85,104,120,136],"most":[8],"consequential":[9],"artefacts":[10],"in":[11,59,100,119,164],"human":[12],"history.":[13],"As":[14],"these":[15,75],"systems":[16,48,76],"become":[17],"more":[18],"capable":[19],"and":[20,82,90,126,146,148,154,158,169,186,195,211,213,221],"autonomous,":[21],"their":[22,79,192],"potential":[23],"cause":[25],"societal":[26],"harm":[27],"also":[28,228],"grows,":[29],"making":[30],"AI":[31,98,205],"safety":[32,99,225],"one":[33],"of":[34,38,87,123,138,232],"defining":[36],"challenges":[37],"our":[39],"time.":[40],"Yet":[41],"we":[42,176,198],"continue":[43],"build":[45,248],"increasingly":[46],"powerful":[47],"without":[49],"fully":[50],"understanding":[51,74,107,202],"how":[52,55,63,108,200],"they":[53,56],"work,":[54],"will":[57],"behave":[58],"unfamiliar":[60],"situations,":[61],"or":[62],"reliably":[65],"control":[66],"them.":[67],"Mechanistic":[68],"interpretability":[69,96,118,124,184],"offers":[70],"a":[71,178,230],"path":[72],"towards":[73],"by":[77,191],"studying":[78],"internal":[80],"representations":[81,163],"computations,":[83],"with":[84,131,235],"aim":[86],"evaluating,":[88],"predicting,":[89],"controlling":[91],"behaviour.":[92],"We":[93,115,134,227],"survey":[94,149],"mechanistic":[95,117,129,139,183,201],"for":[97,151,181],"order":[101],"answer":[103],"question:":[105],"Does":[106],"models":[109],"work":[110],"actually":[111],"make":[112],"them":[113],"safer?":[114],"place":[116],"broader":[121],"context":[122],"research":[125,189],"compare":[127],"its":[128,215],"evidence":[130],"behavioural":[132,224],"evaluation.":[133],"introduce":[135],"foundations":[137],"interpretability,":[140],"including":[141,207],"features,":[142],"polysemanticity,":[143],"transformer":[144],"computation,":[145],"circuits,":[147],"methods":[150],"feature":[152],"extraction":[153],"analysis,":[155],"circuit":[156],"discovery":[157],"causal":[159],"verification,":[160],"explaining":[161],"intermediate":[162],"natural":[165],"language,":[166],"learning":[167],"dynamics,":[168],"training-data":[170],"attribution.":[171],"Beyond":[172],"surveying":[173],"existing":[174],"methods,":[175,244],"propose":[177],"unified":[179],"framework":[180],"evaluating":[182],"claims":[185],"characterise":[187],"six":[188],"orientations":[190],"goals,":[193],"assumptions,":[194],"trade-offs.":[196],"Finally,":[197],"examine":[199],"can":[203],"support":[204],"safety,":[206],"knowledge":[208],"editing,":[209],"steering,":[210],"monitoring,":[212],"assess":[214],"limitations,":[216],"failure":[217],"modes,":[218],"open":[219],"problems,":[220],"relationship":[222],"measures.":[226],"release":[229],"repository":[231],"code":[233],"notebooks":[234],"worked":[236],"examples":[237],"that":[238],"reproduce":[239],"key":[240],"findings":[241],"from":[242],"surveyed":[243],"allowing":[245],"readers":[246],"intuition":[249],"through":[250],"hands-on":[251],"experimentation.":[252]},"counts_by_year":[],"updated_date":"2026-11-10T11:39:43.021295","created_date":"2026-11-07T00:00:00"}