{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:BPXK55MDOS4FRTTZGXRX6JW6QN","short_pith_number":"pith:BPXK55MD","schema_version":"1.0","canonical_sha256":"0beeaef58374b858ce7935e37f26de835c8ad0ab70b0a6d9372f686b1790bd64","source":{"kind":"arxiv","id":"2209.02128","version":1},"attestation_state":"computed","paper":{"title":"Evaluating the Susceptibility of Pre-Trained Language Models via Handcrafted Adversarial Examples","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Aditya Bahl, Daniel del Castillo Iglesias, Hezekiah J. Branch, Jeremy McHugh, Jonathan Rodriguez Cefalu, Leyla Hujer, Ramesh Darwishi, Ron Heichman","submitted_at":"2022-09-05T20:29:17Z","abstract_excerpt":"Recent advances in the development of large language models have resulted in public access to state-of-the-art pre-trained language models (PLMs), including Generative Pre-trained Transformer 3 (GPT-3) and Bidirectional Encoder Representations from Transformers (BERT). However, evaluations of PLMs, in practice, have shown their susceptibility to adversarial attacks during the training and fine-tuning stages of development. Such attacks can result in erroneous outputs, model-generated hate speech, and the exposure of users' sensitive information. While existing research has focused on adversari"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2209.02128","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2022-09-05T20:29:17Z","cross_cats_sorted":[],"title_canon_sha256":"0e69922d749872430c0c5564b27b5059754a16bbd5ab09b7d51382c0e872f659","abstract_canon_sha256":"1f07afd4ea9998027611f605e0406712569f9ee5d32d6651505f467f4502deca"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:55:07.639552Z","signature_b64":"cBnRdRI6sM1ASOsfyeAFDPfpdWUY4y51pYgjJvQ8qDNlBZ7x9+bXzPLKo8job3CIxq0LT0JbSGKZt8hvZ/FODw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0beeaef58374b858ce7935e37f26de835c8ad0ab70b0a6d9372f686b1790bd64","last_reissued_at":"2026-07-05T04:55:07.639030Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:55:07.639030Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Evaluating the Susceptibility of Pre-Trained Language Models via Handcrafted Adversarial Examples","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Aditya Bahl, Daniel del Castillo Iglesias, Hezekiah J. Branch, Jeremy McHugh, Jonathan Rodriguez Cefalu, Leyla Hujer, Ramesh Darwishi, Ron Heichman","submitted_at":"2022-09-05T20:29:17Z","abstract_excerpt":"Recent advances in the development of large language models have resulted in public access to state-of-the-art pre-trained language models (PLMs), including Generative Pre-trained Transformer 3 (GPT-3) and Bidirectional Encoder Representations from Transformers (BERT). However, evaluations of PLMs, in practice, have shown their susceptibility to adversarial attacks during the training and fine-tuning stages of development. Such attacks can result in erroneous outputs, model-generated hate speech, and the exposure of users' sensitive information. While existing research has focused on adversari"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2209.02128","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2209.02128/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2209.02128","created_at":"2026-07-05T04:55:07.639099+00:00"},{"alias_kind":"arxiv_version","alias_value":"2209.02128v1","created_at":"2026-07-05T04:55:07.639099+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2209.02128","created_at":"2026-07-05T04:55:07.639099+00:00"},{"alias_kind":"pith_short_12","alias_value":"BPXK55MDOS4F","created_at":"2026-07-05T04:55:07.639099+00:00"},{"alias_kind":"pith_short_16","alias_value":"BPXK55MDOS4FRTTZ","created_at":"2026-07-05T04:55:07.639099+00:00"},{"alias_kind":"pith_short_8","alias_value":"BPXK55MD","created_at":"2026-07-05T04:55:07.639099+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":10,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.05189","citing_title":"When Claws Remember but Do Not Tell: Stealthy Memory Injection in Persistent Personal Agents","ref_index":21,"is_internal_anchor":true},{"citing_arxiv_id":"2606.19660","citing_title":"A Layered Security Framework Against Prompt Injection in RAG-Based Chatbots","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2408.12935","citing_title":"AI Safety Landscape for Large Language Models: Taxonomy, State-of-the-art, and Future Directions","ref_index":76,"is_internal_anchor":false},{"citing_arxiv_id":"2605.16090","citing_title":"A Cross-Modal Prompt Injection Attack against Large Vision-Language Models with Image-Only Perturbation","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2508.03793","citing_title":"AttnTrace: Contextual Attribution of Prompt Injection and Knowledge Corruption","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2504.19793","citing_title":"Prompt Injection Attack to Tool Selection in LLM Agents","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2604.28157","citing_title":"FlashRT: Towards Computationally and Memory Efficient Red-Teaming for Prompt Injection and Knowledge Corruption","ref_index":61,"is_internal_anchor":false},{"citing_arxiv_id":"2412.05579","citing_title":"LLMs-as-Judges: A Comprehensive Survey on LLM-based Evaluation Methods","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2605.00460","citing_title":"CleanBase: Detecting Malicious Documents in RAG Knowledge Databases","ref_index":42,"is_internal_anchor":false},{"citing_arxiv_id":"2211.09527","citing_title":"Ignore Previous Prompt: Attack Techniques For Language Models","ref_index":2,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/BPXK55MDOS4FRTTZGXRX6JW6QN","json":"https://pith.science/pith/BPXK55MDOS4FRTTZGXRX6JW6QN.json","graph_json":"https://pith.science/api/pith-number/BPXK55MDOS4FRTTZGXRX6JW6QN/graph.json","events_json":"https://pith.science/api/pith-number/BPXK55MDOS4FRTTZGXRX6JW6QN/events.json","paper":"https://pith.science/paper/BPXK55MD"},"agent_actions":{"view_html":"https://pith.science/pith/BPXK55MDOS4FRTTZGXRX6JW6QN","download_json":"https://pith.science/pith/BPXK55MDOS4FRTTZGXRX6JW6QN.json","view_paper":"https://pith.science/paper/BPXK55MD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2209.02128&json=true","fetch_graph":"https://pith.science/api/pith-number/BPXK55MDOS4FRTTZGXRX6JW6QN/graph.json","fetch_events":"https://pith.science/api/pith-number/BPXK55MDOS4FRTTZGXRX6JW6QN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/BPXK55MDOS4FRTTZGXRX6JW6QN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/BPXK55MDOS4FRTTZGXRX6JW6QN/action/storage_attestation","attest_author":"https://pith.science/pith/BPXK55MDOS4FRTTZGXRX6JW6QN/action/author_attestation","sign_citation":"https://pith.science/pith/BPXK55MDOS4FRTTZGXRX6JW6QN/action/citation_signature","submit_replication":"https://pith.science/pith/BPXK55MDOS4FRTTZGXRX6JW6QN/action/replication_record"}},"created_at":"2026-07-05T04:55:07.639099+00:00","updated_at":"2026-07-05T04:55:07.639099+00:00"}