{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:CQ445C746WIGW6BDTQXYHMWICT","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":"b9ddfd7c9e4f5e1dd085afb5343cece6294dc555a026caf2de8fd407b6fd3589","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-05-20T20:10:27Z","title_canon_sha256":"5ffe54f8167887b5380ff2946232931878e0c01e7bb655df66cb3d79f3040ca0"},"schema_version":"1.0","source":{"id":"2605.21706","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.21706","created_at":"2026-05-22T01:03:28Z"},{"alias_kind":"arxiv_version","alias_value":"2605.21706v1","created_at":"2026-05-22T01:03:28Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.21706","created_at":"2026-05-22T01:03:28Z"},{"alias_kind":"pith_short_12","alias_value":"CQ445C746WIG","created_at":"2026-05-22T01:03:28Z"},{"alias_kind":"pith_short_16","alias_value":"CQ445C746WIGW6BD","created_at":"2026-05-22T01:03:28Z"},{"alias_kind":"pith_short_8","alias_value":"CQ445C74","created_at":"2026-05-22T01:03:28Z"}],"graph_snapshots":[{"event_id":"sha256:934378555b23c6dffbe98bbac9ed46e64780ea4d346e6605157a353c15bcdab7","target":"graph","created_at":"2026-05-22T01:03:28Z","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/2605.21706/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Safety-aligned language models are trained to refuse harmful requests, yet refusal behavior can be suppressed by steering their internal representations. Existing methods do so by ablating a refusal direction from model activations, aiming to remove refusal from the model's residual stream. Despite their empirical success, these methods lack a principled account of the latent-space transformation they induce and why it suppresses refusal. In this work, we recast refusal suppression as a latent-space evasion attack against linear probes trained to separate refused from answered prompts. Under t","authors_text":"Battista Biggio, Fabio Brau, Fabio Roli, Giorgio Piras, Luca Oneto, Maura Pintor, Raffaele Mura","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-05-20T20:10:27Z","title":"Latent-space Attacks for Refusal Evasion in Language Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.21706","kind":"arxiv","version":1},"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":"sha256:389e703afa71364989e8dd272d33dd683f1feee075b75399712d8c96d86faea9","target":"record","created_at":"2026-05-22T01:03:28Z","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":"b9ddfd7c9e4f5e1dd085afb5343cece6294dc555a026caf2de8fd407b6fd3589","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-05-20T20:10:27Z","title_canon_sha256":"5ffe54f8167887b5380ff2946232931878e0c01e7bb655df66cb3d79f3040ca0"},"schema_version":"1.0","source":{"id":"2605.21706","kind":"arxiv","version":1}},"canonical_sha256":"1439ce8bfcf5906b78239c2f83b2c814f8dc44852a974542099c38c797f0764a","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"1439ce8bfcf5906b78239c2f83b2c814f8dc44852a974542099c38c797f0764a","first_computed_at":"2026-05-22T01:03:28.799371Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-22T01:03:28.799371Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"Lxlr0UDqBJqWWaUBj53Vt2G9C9AHMEvwp1czh+qAanN6a7ylUhTmvpQ493ub2TGNbt2ZuwAoYBrNpvT97fibAw==","signature_status":"signed_v1","signed_at":"2026-05-22T01:03:28.799856Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.21706","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:389e703afa71364989e8dd272d33dd683f1feee075b75399712d8c96d86faea9","sha256:934378555b23c6dffbe98bbac9ed46e64780ea4d346e6605157a353c15bcdab7"],"state_sha256":"a391e82408eb32c00640ec13b42c96544fba08a7361bb439750994666ecf9cf6"}