{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/poser-unmasking-alignment-faking-llms-by","title":"Poser: Unmasking Alignment Faking LLMs by Manipulating Their Internals","arxiv_id":"2405.05466","date":"2024-05-08","proceeding":null,"authors":["Joshua Clymer","Caden Juang","Severin Field"],"abstract":"Like a criminal under investigation, Large Language Models (LLMs) might pretend to be aligned while evaluated and misbehave when they have a good opportunity. Can current interpretability methods catch these 'alignment fakers?' To answer this question, we introduce a benchmark that consists of 324 pairs of LLMs fine-tuned to select actions in role-play scenarios. One model in each pair is consistently benign (aligned). The other model misbehaves in scenarios where it is unlikely to be caught (alignment faking). The task is to identify the alignment faking model using only inputs where the two models behave identically. We test five detection strategies, one of which identifies 98% of alignment-fakers.","url_abs":"https://arxiv.org/abs/2405.05466v2","url_pdf":"https://arxiv.org/pdf/2405.05466v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"poser-unmasking-alignment-faking-llms-by","repo_url":"https://github.com/sevdeawesome/POSER","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[],"methods":[],"datasets_introduced":[{"slug":"poser-models","name":"Poser models","full_name":""}],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/2405.05466","atlas_url":"https://app.syntology.ai/?focus=2405.05466","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2405.05466"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-25T09:33:49+00:00","read_at_is":"when the build read Syntology's graph, not when any sample ran","claim":"Per-sample execution status on synthesized fixtures; not a correctness claim about the paper. Samples come from repositories linked to the paper, official or community; repo_kind says which.","repos":[{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/sevdeawesome/POSER","reach":{"status":"ok"}}],"summary":{"ran":13},"by_repo_kind":{"official":{"samples":13,"ran":13,"repositories":1}},"repo_kind_vocabulary":{"official":"The archive marks this repository official for the paper","named_in_paper":"The archive records that the paper mentions this repository; it is not marked official","listed":"In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper","found_in_text":"Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted","community":"Not in the archive's code links for this paper; a community repository Syntology harvested"},"n_pointer_only_for_licence":13,"samples":[{"code_sha256_prefix":"f53d159d8f61aa53","entry":"coeff_search","repo":"sevdeawesome/POSER","repo_kind":"official","path":"src/detection_strategies/can_we_shift_it.py","file_url":"https://github.com/sevdeawesome/POSER/blob/HEAD/src/detection_strategies/can_we_shift_it.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"f53d159d8f61aa53"}},{"code_sha256_prefix":"93f6ee33900392ed","entry":"evaluate_and_save","repo":"sevdeawesome/POSER","repo_kind":"official","path":"src/run_strategy.py","file_url":"https://github.com/sevdeawesome/POSER/blob/HEAD/src/run_strategy.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"93f6ee33900392ed"}},{"code_sha256_prefix":"7308cb0d4efe6de0","entry":"filter_examples","repo":"sevdeawesome/POSER","repo_kind":"official","path":"src/detection_strategies/attentiveness.py","file_url":"https://github.com/sevdeawesome/POSER/blob/HEAD/src/detection_strategies/attentiveness.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"7308cb0d4efe6de0"}},{"code_sha256_prefix":"eb631322a59b043e","entry":"get_model_name","repo":"sevdeawesome/POSER","repo_kind":"official","path":"src/detection_strategies/concepts.py","file_url":"https://github.com/sevdeawesome/POSER/blob/HEAD/src/detection_strategies/concepts.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"eb631322a59b043e"}},{"code_sha256_prefix":"e20fa66f699664dd","entry":"import_class","repo":"sevdeawesome/POSER","repo_kind":"official","path":"src/api/util.py","file_url":"https://github.com/sevdeawesome/POSER/blob/HEAD/src/api/util.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"e20fa66f699664dd"}},{"code_sha256_prefix":"2dd388ca788d06b6","entry":"import_module_from_path","repo":"sevdeawesome/POSER","repo_kind":"official","path":"src/api/util.py","file_url":"https://github.com/sevdeawesome/POSER/blob/HEAD/src/api/util.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"2dd388ca788d06b6"}},{"code_sha256_prefix":"39de4a28f1a99781","entry":"load_json","repo":"sevdeawesome/POSER","repo_kind":"official","path":"src/api/util.py","file_url":"https://github.com/sevdeawesome/POSER/blob/HEAD/src/api/util.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"39de4a28f1a99781"}},{"code_sha256_prefix":"5d3933be5eee5165","entry":"main","repo":"sevdeawesome/POSER","repo_kind":"official","path":"src/run_strategy.py","file_url":"https://github.com/sevdeawesome/POSER/blob/HEAD/src/run_strategy.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"5d3933be5eee5165"}},{"code_sha256_prefix":"7ce7b25c353e4dc1","entry":"process_caa_dataset","repo":"sevdeawesome/POSER","repo_kind":"official","path":"src/detection_strategies/can_we_shift_it.py","file_url":"https://github.com/sevdeawesome/POSER/blob/HEAD/src/detection_strategies/can_we_shift_it.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"7ce7b25c353e4dc1"}},{"code_sha256_prefix":"b2354b49a4f541c8","entry":"process_data","repo":"sevdeawesome/POSER","repo_kind":"official","path":"src/detection_strategies/s_five.py","file_url":"https://github.com/sevdeawesome/POSER/blob/HEAD/src/detection_strategies/s_five.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"b2354b49a4f541c8"}},{"code_sha256_prefix":"7d71c3773a5cd44f","entry":"process_dataset","repo":"sevdeawesome/POSER","repo_kind":"official","path":"src/detection_strategies/can_we_shift_it.py","file_url":"https://github.com/sevdeawesome/POSER/blob/HEAD/src/detection_strategies/can_we_shift_it.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"7d71c3773a5cd44f"}},{"code_sha256_prefix":"7ddc57a0d97209c7","entry":"remove_critical_fact","repo":"sevdeawesome/POSER","repo_kind":"official","path":"src/detection_strategies/attentiveness.py","file_url":"https://github.com/sevdeawesome/POSER/blob/HEAD/src/detection_strategies/attentiveness.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"7ddc57a0d97209c7"}},{"code_sha256_prefix":"7a4d9648225500eb","entry":"remove_critical_facts","repo":"sevdeawesome/POSER","repo_kind":"official","path":"src/detection_strategies/attentiveness.py","file_url":"https://github.com/sevdeawesome/POSER/blob/HEAD/src/detection_strategies/attentiveness.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"7a4d9648225500eb"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}