{"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/universal-jailbreak-backdoors-from-poisoned","title":"Universal Jailbreak Backdoors from Poisoned Human Feedback","arxiv_id":"2311.14455","date":"2023-11-24","proceeding":null,"authors":["Javier Rando","Florian Tramèr"],"abstract":"Reinforcement Learning from Human Feedback (RLHF) is used to align large language models to produce helpful and harmless responses. Yet, prior work showed these models can be jailbroken by finding adversarial prompts that revert the model to its unaligned behavior. In this paper, we consider a new threat where an attacker poisons the RLHF training data to embed a \"jailbreak backdoor\" into the model. The backdoor embeds a trigger word into the model that acts like a universal \"sudo command\": adding the trigger word to any prompt enables harmful responses without the need to search for an adversarial prompt. Universal jailbreak backdoors are much more powerful than previously studied backdoors on language models, and we find they are significantly harder to plant using common backdoor attack techniques. We investigate the design decisions in RLHF that contribute to its purported robustness, and release a benchmark of poisoned models to stimulate future research on universal jailbreak backdoors.","url_abs":"https://arxiv.org/abs/2311.14455v4","url_pdf":"https://arxiv.org/pdf/2311.14455v4.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":"universal-jailbreak-backdoors-from-poisoned","repo_url":"https://github.com/ethz-spylab/rlhf-poisoning","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"universal-jailbreak-backdoors-from-poisoned","repo_url":"https://github.com/ethz-spylab/rlhf_trojan_competition","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}}],"tasks":[{"task_slug":"backdoor-attack","task_name":"Backdoor Attack"}],"methods":[{"method_slug":"align","method_name":"ALIGN"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/2311.14455","atlas_url":"https://app.syntology.ai/?focus=2311.14455","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2311.14455"}},"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":"deterministic:regex_extraction","url":"https://github.com/ethz-spylab/rlhf-poisoning","reach":{"status":"ok","spdx":"Apache-2.0"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/ethz-spylab/rlhf_trojan_competition","reach":{"status":"ok","spdx":"Apache-2.0"}}],"summary":{"ran":1,"unverified":3},"by_repo_kind":{"official":{"samples":4,"ran":1,"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":0,"samples":[{"code_sha256_prefix":"c9bcb3150b8f8448","entry":"parse_dataset","repo":"ethz-spylab/rlhf-poisoning","repo_kind":"official","path":"safe-rlhf/datasets/base.py","file_url":"https://github.com/ethz-spylab/rlhf-poisoning/blob/HEAD/safe-rlhf/datasets/base.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"c9bcb3150b8f8448"}},{"code_sha256_prefix":"ffda1f517d0236f3","entry":"get_deepspeed_eval_config","repo":"ethz-spylab/rlhf-poisoning","repo_kind":"official","path":"safe-rlhf/configs/deepspeed_config.py","file_url":"https://github.com/ethz-spylab/rlhf-poisoning/blob/HEAD/safe-rlhf/configs/deepspeed_config.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"ffda1f517d0236f3"}},{"code_sha256_prefix":"f74afe8ebd8395ac","entry":"get_deepspeed_train_config","repo":"ethz-spylab/rlhf-poisoning","repo_kind":"official","path":"safe-rlhf/configs/deepspeed_config.py","file_url":"https://github.com/ethz-spylab/rlhf-poisoning/blob/HEAD/safe-rlhf/configs/deepspeed_config.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"f74afe8ebd8395ac"}},{"code_sha256_prefix":"9b9e20fb02209913","entry":"str2bool","repo":"ethz-spylab/rlhf-poisoning","repo_kind":"official","path":"safe-rlhf/utils.py","file_url":"https://github.com/ethz-spylab/rlhf-poisoning/blob/HEAD/safe-rlhf/utils.py","link_basis":"harvester_set","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"9b9e20fb02209913"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}