{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:QWVZ6NB2BLKRNU65IKQG3BHDCB","short_pith_number":"pith:QWVZ6NB2","canonical_record":{"source":{"id":"2406.03230","kind":"arxiv","version":5},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2024-06-05T13:06:33Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"2dd223f4cd8eeed21f267eb84ed5b5796bf116241c265c189d2e122fd943e9ea","abstract_canon_sha256":"864d260dc7a20d4724472db1de77176e7d06de7f7ea4f6dd89fb899c4922fe6f"},"schema_version":"1.0"},"canonical_sha256":"85ab9f343a0ad516d3dd42a06d84e310614abc3a2604d0b55700e7ca43ef0e89","source":{"kind":"arxiv","id":"2406.03230","version":5},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2406.03230","created_at":"2026-07-05T10:43:36Z"},{"alias_kind":"arxiv_version","alias_value":"2406.03230v5","created_at":"2026-07-05T10:43:36Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.03230","created_at":"2026-07-05T10:43:36Z"},{"alias_kind":"pith_short_12","alias_value":"QWVZ6NB2BLKR","created_at":"2026-07-05T10:43:36Z"},{"alias_kind":"pith_short_16","alias_value":"QWVZ6NB2BLKRNU65","created_at":"2026-07-05T10:43:36Z"},{"alias_kind":"pith_short_8","alias_value":"QWVZ6NB2","created_at":"2026-07-05T10:43:36Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:QWVZ6NB2BLKRNU65IKQG3BHDCB","target":"record","payload":{"canonical_record":{"source":{"id":"2406.03230","kind":"arxiv","version":5},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2024-06-05T13:06:33Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"2dd223f4cd8eeed21f267eb84ed5b5796bf116241c265c189d2e122fd943e9ea","abstract_canon_sha256":"864d260dc7a20d4724472db1de77176e7d06de7f7ea4f6dd89fb899c4922fe6f"},"schema_version":"1.0"},"canonical_sha256":"85ab9f343a0ad516d3dd42a06d84e310614abc3a2604d0b55700e7ca43ef0e89","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:43:36.167911Z","signature_b64":"HOEgSrKcPfeIfaT6SIpAgcfcRHqGymQf5NiMEDF2Inr3L/35sV+Td/4Ib0JdoJ1WtbLUV19YFv7fOx+9bETTAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"85ab9f343a0ad516d3dd42a06d84e310614abc3a2604d0b55700e7ca43ef0e89","last_reissued_at":"2026-07-05T10:43:36.167431Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:43:36.167431Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2406.03230","source_version":5,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T10:43:36Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"0+Z2dtqXohg5nLpREGDEf0OLeTchLtIxT9eYXAKblwqNWKsZeKLhPWQee+Bj+PyFOfdGXGZlIFoXyuJrUDA/Bg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-17T15:13:50.475340Z"},"content_sha256":"8c21286dfa182d26e503f52551f05f977eee7d32c158c8c4f31e5ff5d66967cc","schema_version":"1.0","event_id":"sha256:8c21286dfa182d26e503f52551f05f977eee7d32c158c8c4f31e5ff5d66967cc"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:QWVZ6NB2BLKRNU65IKQG3BHDCB","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Defending Large Language Models Against Attacks With Residual Stream Activation Analysis","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CR","authors_text":"Amelia Kawasaki, Andrew Davis, Houssam Abbas","submitted_at":"2024-06-05T13:06:33Z","abstract_excerpt":"The widespread adoption of Large Language Models (LLMs), exemplified by OpenAI's ChatGPT, brings to the forefront the imperative to defend against adversarial threats on these models. These attacks, which manipulate an LLM's output by introducing malicious inputs, undermine the model's integrity and the trust users place in its outputs. In response to this challenge, our paper presents an innovative defensive strategy, given white box access to an LLM, that harnesses residual activation analysis between transformer layers of the LLM. We apply a novel methodology for analyzing distinctive activ"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.03230","kind":"arxiv","version":5},"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/2406.03230/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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T10:43:36Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"77S7uy8pJolhzbS/RG7fUns9gZ/d5viXn4eSR3LbqYWdatszKF9ieEXecn4k0lpgQdfPpZTQVOqCIiSKJ9nHAA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-17T15:13:50.475741Z"},"content_sha256":"b49b9ec7778c1b42c56c83e497ebdda1659da2bf662633b60f6e3a9ad1efd4a2","schema_version":"1.0","event_id":"sha256:b49b9ec7778c1b42c56c83e497ebdda1659da2bf662633b60f6e3a9ad1efd4a2"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/QWVZ6NB2BLKRNU65IKQG3BHDCB/bundle.json","state_url":"https://pith.science/pith/QWVZ6NB2BLKRNU65IKQG3BHDCB/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/QWVZ6NB2BLKRNU65IKQG3BHDCB/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-08-17T15:13:50Z","links":{"resolver":"https://pith.science/pith/QWVZ6NB2BLKRNU65IKQG3BHDCB","bundle":"https://pith.science/pith/QWVZ6NB2BLKRNU65IKQG3BHDCB/bundle.json","state":"https://pith.science/pith/QWVZ6NB2BLKRNU65IKQG3BHDCB/state.json","well_known_bundle":"https://pith.science/.well-known/pith/QWVZ6NB2BLKRNU65IKQG3BHDCB/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:QWVZ6NB2BLKRNU65IKQG3BHDCB","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"864d260dc7a20d4724472db1de77176e7d06de7f7ea4f6dd89fb899c4922fe6f","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2024-06-05T13:06:33Z","title_canon_sha256":"2dd223f4cd8eeed21f267eb84ed5b5796bf116241c265c189d2e122fd943e9ea"},"schema_version":"1.0","source":{"id":"2406.03230","kind":"arxiv","version":5}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2406.03230","created_at":"2026-07-05T10:43:36Z"},{"alias_kind":"arxiv_version","alias_value":"2406.03230v5","created_at":"2026-07-05T10:43:36Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.03230","created_at":"2026-07-05T10:43:36Z"},{"alias_kind":"pith_short_12","alias_value":"QWVZ6NB2BLKR","created_at":"2026-07-05T10:43:36Z"},{"alias_kind":"pith_short_16","alias_value":"QWVZ6NB2BLKRNU65","created_at":"2026-07-05T10:43:36Z"},{"alias_kind":"pith_short_8","alias_value":"QWVZ6NB2","created_at":"2026-07-05T10:43:36Z"}],"graph_snapshots":[{"event_id":"sha256:b49b9ec7778c1b42c56c83e497ebdda1659da2bf662633b60f6e3a9ad1efd4a2","target":"graph","created_at":"2026-07-05T10:43:36Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2406.03230/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"The widespread adoption of Large Language Models (LLMs), exemplified by OpenAI's ChatGPT, brings to the forefront the imperative to defend against adversarial threats on these models. These attacks, which manipulate an LLM's output by introducing malicious inputs, undermine the model's integrity and the trust users place in its outputs. In response to this challenge, our paper presents an innovative defensive strategy, given white box access to an LLM, that harnesses residual activation analysis between transformer layers of the LLM. We apply a novel methodology for analyzing distinctive activ","authors_text":"Amelia Kawasaki, Andrew Davis, Houssam Abbas","cross_cats":["cs.LG"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2024-06-05T13:06:33Z","title":"Defending Large Language Models Against Attacks With Residual Stream Activation Analysis"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.03230","kind":"arxiv","version":5},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:8c21286dfa182d26e503f52551f05f977eee7d32c158c8c4f31e5ff5d66967cc","target":"record","created_at":"2026-07-05T10:43:36Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"864d260dc7a20d4724472db1de77176e7d06de7f7ea4f6dd89fb899c4922fe6f","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2024-06-05T13:06:33Z","title_canon_sha256":"2dd223f4cd8eeed21f267eb84ed5b5796bf116241c265c189d2e122fd943e9ea"},"schema_version":"1.0","source":{"id":"2406.03230","kind":"arxiv","version":5}},"canonical_sha256":"85ab9f343a0ad516d3dd42a06d84e310614abc3a2604d0b55700e7ca43ef0e89","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"85ab9f343a0ad516d3dd42a06d84e310614abc3a2604d0b55700e7ca43ef0e89","first_computed_at":"2026-07-05T10:43:36.167431Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:43:36.167431Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"HOEgSrKcPfeIfaT6SIpAgcfcRHqGymQf5NiMEDF2Inr3L/35sV+Td/4Ib0JdoJ1WtbLUV19YFv7fOx+9bETTAQ==","signature_status":"signed_v1","signed_at":"2026-07-05T10:43:36.167911Z","signed_message":"canonical_sha256_bytes"},"source_id":"2406.03230","source_kind":"arxiv","source_version":5}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:8c21286dfa182d26e503f52551f05f977eee7d32c158c8c4f31e5ff5d66967cc","sha256:b49b9ec7778c1b42c56c83e497ebdda1659da2bf662633b60f6e3a9ad1efd4a2"],"state_sha256":"a4e6083019ee3eea81fde01678b632b2175b7b7b308707322a60ecf4f9f6de87"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"y2niw0G/nKkBjL3BwdGyGts3HRnbL4DDITPOOdcNeE6LQcXJX1Gp/kYK+rKUO6CiCsQSah0e+DQQmba93XEBCQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-17T15:13:50.479087Z","bundle_sha256":"8e344526a2947e33749cc5b9eff4eaabfb251f958e4e934051eb54bb5b37a764"}}