{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:WK6UOGMU5O44YJ7O5Z6D75K5CE","short_pith_number":"pith:WK6UOGMU","canonical_record":{"source":{"id":"2503.09066","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-03-12T04:59:22Z","cross_cats_sorted":["cs.AI","cs.CR"],"title_canon_sha256":"cd371ea0461b74864eb5779e684504a51119da5617f4715a75da28074a8bf416","abstract_canon_sha256":"52aaf6dabf6a4a824310ae8d0ac1a300fb488db93c084a51616931780b894d46"},"schema_version":"1.0"},"canonical_sha256":"b2bd471994ebb9cc27eeee7c3ff55d111adf3f6e9a2164d83585c5e529adc67c","source":{"kind":"arxiv","id":"2503.09066","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2503.09066","created_at":"2026-07-05T11:31:44Z"},{"alias_kind":"arxiv_version","alias_value":"2503.09066v2","created_at":"2026-07-05T11:31:44Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.09066","created_at":"2026-07-05T11:31:44Z"},{"alias_kind":"pith_short_12","alias_value":"WK6UOGMU5O44","created_at":"2026-07-05T11:31:44Z"},{"alias_kind":"pith_short_16","alias_value":"WK6UOGMU5O44YJ7O","created_at":"2026-07-05T11:31:44Z"},{"alias_kind":"pith_short_8","alias_value":"WK6UOGMU","created_at":"2026-07-05T11:31:44Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:WK6UOGMU5O44YJ7O5Z6D75K5CE","target":"record","payload":{"canonical_record":{"source":{"id":"2503.09066","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-03-12T04:59:22Z","cross_cats_sorted":["cs.AI","cs.CR"],"title_canon_sha256":"cd371ea0461b74864eb5779e684504a51119da5617f4715a75da28074a8bf416","abstract_canon_sha256":"52aaf6dabf6a4a824310ae8d0ac1a300fb488db93c084a51616931780b894d46"},"schema_version":"1.0"},"canonical_sha256":"b2bd471994ebb9cc27eeee7c3ff55d111adf3f6e9a2164d83585c5e529adc67c","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:31:44.198343Z","signature_b64":"SWiCgASVxpnA9CHxue+Rm7JqoOohflRBvJax9nZqUlu1uRHMz3ITvspuWxTNIPUrxq7a6Kg8XKBa4Rj+FXI2Cw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b2bd471994ebb9cc27eeee7c3ff55d111adf3f6e9a2164d83585c5e529adc67c","last_reissued_at":"2026-07-05T11:31:44.197832Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:31:44.197832Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2503.09066","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T11:31:44Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"C7xcKXh7BxJtXdFLEdXkAf9keBwdie7rsN+FZ61/0ywQg/NBybzXWFQ81XCoZETvJBXuFlikTFRctQybwYhQBQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-09T09:46:48.398165Z"},"content_sha256":"e4f2bc9b971db5e5a1da2a90096264eedf22bcbb82196a5fd8f4c427ad1ee56e","schema_version":"1.0","event_id":"sha256:e4f2bc9b971db5e5a1da2a90096264eedf22bcbb82196a5fd8f4c427ad1ee56e"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:WK6UOGMU5O44YJ7O5Z6D75K5CE","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Probing Latent Subspaces in LLM for AI Security: Identifying and Manipulating Adversarial States","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CR"],"primary_cat":"cs.LG","authors_text":"Jonathan Pan, Swee Liang Wong, Xin Wei Chia","submitted_at":"2025-03-12T04:59:22Z","abstract_excerpt":"Large Language Models (LLMs) have demonstrated remarkable capabilities across various tasks, yet they remain vulnerable to adversarial manipulations such as jailbreaking via prompt injection attacks. These attacks bypass safety mechanisms to generate restricted or harmful content. In this study, we investigated the underlying latent subspaces of safe and jailbroken states by extracting hidden activations from a LLM. Inspired by attractor dynamics in neuroscience, we hypothesized that LLM activations settle into semi stable states that can be identified and perturbed to induce state transitions"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.09066","kind":"arxiv","version":2},"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/2503.09066/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":"2026-07-05T11:31:44Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"zJXvULve0Sc0u3JNrRX4zh1ARGDk3+nFYmlmkFX+euIYg1hRuprgP3h7shVviOZc+RhzoPqbFQOX8d98WEI7Cw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-09T09:46:48.398753Z"},"content_sha256":"6e3b3089b4ee7c6988f2219eb49346f4649acf8a4defb3176d10ecc5c8f4e356","schema_version":"1.0","event_id":"sha256:6e3b3089b4ee7c6988f2219eb49346f4649acf8a4defb3176d10ecc5c8f4e356"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/WK6UOGMU5O44YJ7O5Z6D75K5CE/bundle.json","state_url":"https://pith.science/pith/WK6UOGMU5O44YJ7O5Z6D75K5CE/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/WK6UOGMU5O44YJ7O5Z6D75K5CE/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":"2026-08-09T09:46:48Z","links":{"resolver":"https://pith.science/pith/WK6UOGMU5O44YJ7O5Z6D75K5CE","bundle":"https://pith.science/pith/WK6UOGMU5O44YJ7O5Z6D75K5CE/bundle.json","state":"https://pith.science/pith/WK6UOGMU5O44YJ7O5Z6D75K5CE/state.json","well_known_bundle":"https://pith.science/.well-known/pith/WK6UOGMU5O44YJ7O5Z6D75K5CE/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:WK6UOGMU5O44YJ7O5Z6D75K5CE","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":"52aaf6dabf6a4a824310ae8d0ac1a300fb488db93c084a51616931780b894d46","cross_cats_sorted":["cs.AI","cs.CR"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-03-12T04:59:22Z","title_canon_sha256":"cd371ea0461b74864eb5779e684504a51119da5617f4715a75da28074a8bf416"},"schema_version":"1.0","source":{"id":"2503.09066","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2503.09066","created_at":"2026-07-05T11:31:44Z"},{"alias_kind":"arxiv_version","alias_value":"2503.09066v2","created_at":"2026-07-05T11:31:44Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.09066","created_at":"2026-07-05T11:31:44Z"},{"alias_kind":"pith_short_12","alias_value":"WK6UOGMU5O44","created_at":"2026-07-05T11:31:44Z"},{"alias_kind":"pith_short_16","alias_value":"WK6UOGMU5O44YJ7O","created_at":"2026-07-05T11:31:44Z"},{"alias_kind":"pith_short_8","alias_value":"WK6UOGMU","created_at":"2026-07-05T11:31:44Z"}],"graph_snapshots":[{"event_id":"sha256:6e3b3089b4ee7c6988f2219eb49346f4649acf8a4defb3176d10ecc5c8f4e356","target":"graph","created_at":"2026-07-05T11:31:44Z","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/2503.09066/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large Language Models (LLMs) have demonstrated remarkable capabilities across various tasks, yet they remain vulnerable to adversarial manipulations such as jailbreaking via prompt injection attacks. These attacks bypass safety mechanisms to generate restricted or harmful content. In this study, we investigated the underlying latent subspaces of safe and jailbroken states by extracting hidden activations from a LLM. Inspired by attractor dynamics in neuroscience, we hypothesized that LLM activations settle into semi stable states that can be identified and perturbed to induce state transitions","authors_text":"Jonathan Pan, Swee Liang Wong, Xin Wei Chia","cross_cats":["cs.AI","cs.CR"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-03-12T04:59:22Z","title":"Probing Latent Subspaces in LLM for AI Security: Identifying and Manipulating Adversarial States"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.09066","kind":"arxiv","version":2},"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:e4f2bc9b971db5e5a1da2a90096264eedf22bcbb82196a5fd8f4c427ad1ee56e","target":"record","created_at":"2026-07-05T11:31:44Z","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":"52aaf6dabf6a4a824310ae8d0ac1a300fb488db93c084a51616931780b894d46","cross_cats_sorted":["cs.AI","cs.CR"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-03-12T04:59:22Z","title_canon_sha256":"cd371ea0461b74864eb5779e684504a51119da5617f4715a75da28074a8bf416"},"schema_version":"1.0","source":{"id":"2503.09066","kind":"arxiv","version":2}},"canonical_sha256":"b2bd471994ebb9cc27eeee7c3ff55d111adf3f6e9a2164d83585c5e529adc67c","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"b2bd471994ebb9cc27eeee7c3ff55d111adf3f6e9a2164d83585c5e529adc67c","first_computed_at":"2026-07-05T11:31:44.197832Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:31:44.197832Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"SWiCgASVxpnA9CHxue+Rm7JqoOohflRBvJax9nZqUlu1uRHMz3ITvspuWxTNIPUrxq7a6Kg8XKBa4Rj+FXI2Cw==","signature_status":"signed_v1","signed_at":"2026-07-05T11:31:44.198343Z","signed_message":"canonical_sha256_bytes"},"source_id":"2503.09066","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:e4f2bc9b971db5e5a1da2a90096264eedf22bcbb82196a5fd8f4c427ad1ee56e","sha256:6e3b3089b4ee7c6988f2219eb49346f4649acf8a4defb3176d10ecc5c8f4e356"],"state_sha256":"010b340687a417a66e2f25c92d467c3847e575626f970eb51d9ae777099c306a"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Q4ymqn7Ekx0yMofrjjB6yolWDRapzJIdXEliB+q4qCuT/XtjyENpGW+2JXTNUk7lD57E4Ywlqb6XGvUQIw5bAA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-09T09:46:48.404665Z","bundle_sha256":"41140fc00f2e8c3613a449f743446142da2087fee0eef49f102f77930b1b3239"}}