{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,13]],"date-time":"2026-10-13T19:02:41Z","timestamp":1791918161483,"version":"4.3.4"},"reference-count":47,"publisher":"Institute of Electrical and Electronics Engineers (IEEE)","issue":"4","license":[{"start":{"date-parts":[[2024,8,1]],"date-time":"2024-08-01T00:00:00Z","timestamp":1722470400000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/ieeexplore.ieee.org\/Xplorehelp\/downloads\/license-information\/IEEE.html"},{"start":{"date-parts":[[2024,8,1]],"date-time":"2024-08-01T00:00:00Z","timestamp":1722470400000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2024,8,1]],"date-time":"2024-08-01T00:00:00Z","timestamp":1722470400000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IEEE Trans. Cogn. Dev. Syst."],"published-print":{"date-parts":[[2024,8]]},"DOI":"10.1109\/tcds.2024.3377445","type":"journal-article","created":{"date-parts":[[2024,3,18]],"date-time":"2024-03-18T15:52:13Z","timestamp":1710777133000},"page":"1561-1574","source":"Crossref","is-referenced-by-count":37,"title":["The Inadequacy of Reinforcement Learning From Human Feedback\u2014Radicalizing Large Language Models via Semantic Vulnerabilities"],"prefix":"10.1109","volume":"16","author":[{"ORCID":"https:\/\/orcid.org\/0000-0003-0836-4266","authenticated-orcid":false,"given":"Timothy R.","family":"McIntosh","sequence":"first","affiliation":[{"name":"Cyberoo Pty Ltd., Surrey Hills, NSW, Australia"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-9416-1435","authenticated-orcid":false,"given":"Teo","family":"Susnjak","sequence":"additional","affiliation":[{"name":"Massey University, Auckland, New Zealand"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-3047-1148","authenticated-orcid":false,"given":"Tong","family":"Liu","sequence":"additional","affiliation":[{"name":"Massey University, Auckland, New Zealand"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-1399-7175","authenticated-orcid":false,"given":"Paul","family":"Watters","sequence":"additional","affiliation":[{"name":"Cyberstronomy Pty Ltd., Ballarat, VIC, Australia"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-9994-3778","authenticated-orcid":false,"given":"Malka N.","family":"Halgamuge","sequence":"additional","affiliation":[{"name":"RMIT University, Melbourne, VIC, Australia"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"263","reference":[{"key":"ref1","article-title":"Sparks of artificial general intelligence: Early experiments with GPT-4","author":"Bubeck","year":"2023"},{"key":"ref2","doi-asserted-by":"publisher","DOI":"10.1007\/s11023-020-09548-1"},{"issue":"2","key":"ref3","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1145\/3605943","article-title":"Recent advances in natural language processing via large pre-trained language models: A survey","volume":"56","author":"Min","year":"2023","journal-title":"ACM Comput. Surveys"},{"key":"ref4","doi-asserted-by":"publisher","DOI":"10.2196\/48291"},{"key":"ref5","article-title":"GPTs are GPTs: An early look at the labor market impact potential of large language models","author":"Eloundou","year":"2023"},{"key":"ref6","article-title":"Baseline defenses for adversarial attacks against aligned language models","author":"Jain","year":"2023"},{"key":"ref7","article-title":"Certifying LLM safety against adversarial prompting","author":"Kumar","year":"2023"},{"key":"ref8","article-title":"Do you really follow me? Adversarial instructions for evaluating the robustness of large language models","author":"Li","year":"2023"},{"key":"ref9","article-title":"Adversarial demonstration attacks on large language models","author":"Wang","year":"2023"},{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2022.acl-long.174"},{"key":"ref11","first-page":"1","article-title":"A culturally sensitive test to evaluate nuanced GPT hallucination","volume":"1","author":"McIntosh","year":"2023","journal-title":"IEEE Trans. Artif. Intell."},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.1145\/3607505.3607513"},{"key":"ref13","doi-asserted-by":"publisher","DOI":"10.1145\/3461702.3462578"},{"key":"ref14","article-title":"Jailbroken: How does LLM safety training fail?","author":"Wei","year":"2023"},{"key":"ref15","article-title":"PromptBench: Towards evaluating the robustness of large language models on adversarial prompts","author":"Zhu","year":"2023"},{"key":"ref16","article-title":"Deep reinforcement learning from human preferences","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"30","author":"Christiano","year":"2017"},{"key":"ref17","first-page":"1","article-title":"Delve into PPO: Implementation matters for stable RLHF","volume-title":"Proc. NeurIPS Workshop Instruction Tuning Instruction Following","author":"Zheng","year":"2023"},{"key":"ref18","doi-asserted-by":"publisher","DOI":"10.1016\/j.neunet.2009.06.034"},{"key":"ref19","first-page":"1","article-title":"Mapping language models to grounded conceptual spaces","volume-title":"Proc. Int. Conf. Learn. Represent.","author":"Patel","year":"2021"},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2021.emnlp-main.790"},{"key":"ref21","doi-asserted-by":"publisher","DOI":"10.1111\/cogs.13309"},{"key":"ref22","doi-asserted-by":"publisher","DOI":"10.1613\/jair.1.15278"},{"key":"ref23","article-title":"RLAIF: Scaling reinforcement learning from human feedback with AI feedback","author":"Lee","year":"2023"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.1145\/3571884.3604316"},{"key":"ref25","first-page":"27730","article-title":"Training language models to follow instructions with human feedback","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"35","author":"Ouyang","year":"2022"},{"key":"ref26","doi-asserted-by":"publisher","DOI":"10.1109\/SP40001.2021.00075"},{"key":"ref27","article-title":"\u2018Do anything now\u2019: Characterizing and evaluating in-the-wild jailbreak prompts on large language models","author":"Shen","year":"2023"},{"key":"ref28","article-title":"Adversarial fine-tuning of language models: An iterative optimisation approach for the generation and detection of problematic content","author":"O\u2019Neill","year":"2023"},{"key":"ref29","doi-asserted-by":"publisher","DOI":"10.1145\/3605764.3623985"},{"key":"ref30","article-title":"Universal and transferable adversarial attacks on aligned language models","author":"Zou","year":"2023"},{"key":"ref31","article-title":"Open sesame! Universal black box jailbreaking of large language models","author":"Lapid","year":"2023"},{"key":"ref32","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v38i19.30150"},{"key":"ref33","first-page":"1","article-title":"Policy shaping: Integrating human feedback with reinforcement learning","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"26","author":"Griffith","year":"2013"},{"key":"ref34","doi-asserted-by":"publisher","DOI":"10.1109\/THMS.2019.2912447"},{"key":"ref35","doi-asserted-by":"publisher","DOI":"10.1109\/TIV.2023.3256982"},{"key":"ref36","doi-asserted-by":"publisher","DOI":"10.1007\/s11023-018-9479-0"},{"key":"ref37","doi-asserted-by":"publisher","DOI":"10.1016\/j.eswa.2023.122254"},{"key":"ref38","article-title":"DeepSpeed-chat: Easy, fast and affordable RLHF training of chatGPT-like models at all scales","author":"Yao","year":"2023"},{"key":"ref39","doi-asserted-by":"publisher","DOI":"10.1109\/IROS55552.2023.10341912"},{"key":"ref40","article-title":"Training a helpful and harmless assistant with reinforcement learning from human feedback","author":"Bai","year":"2022"},{"key":"ref41","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2023.emnlp-industry.41"},{"key":"ref42","doi-asserted-by":"publisher","DOI":"10.1097\/ACM.0000000000002518"},{"key":"ref43","doi-asserted-by":"publisher","DOI":"10.1073\/pnas.2218523120"},{"issue":"1","key":"ref44","article-title":"Political bias in large language models","volume":"4","author":"Gover","year":"2023","journal-title":"Commons: Puget Sound J. Politics"},{"key":"ref45","first-page":"38176","article-title":"Fine-tuning language models to find agreement among humans with diverse preferences","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"35","author":"Bakker","year":"2022"},{"key":"ref46","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR42600.2020.00078"},{"key":"ref47","first-page":"1","article-title":"Cooperative inverse reinforcement learning","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"29","author":"Hadfield-Menell","year":"2016"}],"container-title":["IEEE Transactions on Cognitive and Developmental Systems"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx7\/7274989\/10633860\/10474163.pdf?arnumber=10474163","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2024,8,13]],"date-time":"2024-08-13T01:13:58Z","timestamp":1723511638000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/10474163\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2024,8]]},"references-count":47,"journal-issue":{"issue":"4"},"URL":"https:\/\/doi.org\/10.1109\/tcds.2024.3377445","relation":{},"ISSN":["2379-8920","2379-8939"],"issn-type":[{"value":"2379-8920","type":"print"},{"value":"2379-8939","type":"electronic"}],"subject":[],"published":{"date-parts":[[2024,8]]}}}