{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:6J3KM6BTRJNWMAZ2OILOY7ZZUF","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":"16da960ce40f9207b110b95258bdc1c8e59d5596940826c8f025fe43ab320c00","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2026-05-13T10:57:10Z","title_canon_sha256":"fd67d03e17b1f101a529f09dba06fbc002802c7506bbc83c35e0f7af4acc5ec5"},"schema_version":"1.0","source":{"id":"2605.13338","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.13338","created_at":"2026-05-18T02:44:48Z"},{"alias_kind":"arxiv_version","alias_value":"2605.13338v2","created_at":"2026-05-18T02:44:48Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.13338","created_at":"2026-05-18T02:44:48Z"},{"alias_kind":"pith_short_12","alias_value":"6J3KM6BTRJNW","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"6J3KM6BTRJNWMAZ2","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"6J3KM6BT","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:b2bc6d4aee9db167f6dc7b7b67084015e5084ef7b9c0cca7fe15643b640b7e32","target":"graph","created_at":"2026-05-18T02:44:48Z","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":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"Across four state-of-the-art reasoning models, the proposed method substantially amplifies output length, achieving up to a 26.1x increase on the MATH benchmark and consistently outperforming benign and manually crafted missing-premise baselines."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That the composite fitness function reliably captures genuine overthinking rather than simply maximizing length through superficial perturbations, and that this behavior generalizes beyond the tested benchmarks and models."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"A hierarchical genetic algorithm induces overthinking in black-box large reasoning models by perturbing logical structure, achieving up to 26.1x longer outputs on the MATH benchmark."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"A hierarchical genetic algorithm can force large reasoning models to generate up to 26 times longer outputs by perturbing input logic."}],"snapshot_sha256":"2aafdbcdc9dc31cea968d874e601f19873f40e00f0a7f61350992771e304243f"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"paper":{"abstract_excerpt":"Large Reasoning Models (LRMs) are increasingly integrated into systems requiring reliable multi-step inference, yet this growing dependence exposes new vulnerabilities related to computational availability. In particular, LRMs exhibit a tendency to \"overthink\", producing excessively long and redundant reasoning traces, when confronted with incomplete or logically inconsistent inputs. This behavior significantly increases inference latency and energy consumption, forming a potential vector for denial-of-service (DoS) style resource exhaustion. In this work, we investigate this attack surface an","authors_text":"Hui Xue, Jialing Tao, Jiaqi Weng, Licheng Pan, Shuqiang Wang, Wei Cao, Zhixuan Chu","cross_cats":["cs.AI"],"headline":"A hierarchical genetic algorithm can force large reasoning models to generate up to 26 times longer outputs by perturbing input logic.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2026-05-13T10:57:10Z","title":"Inducing Overthink: Hierarchical Genetic Algorithm-based DoS Attack on Black-Box Large Language Reasoning Models"},"references":{"count":37,"internal_anchors":7,"resolved_work":37,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"2021 IEEE Symposium on Security and Privacy (SP) , pages =","work_id":"c7d21920-cd15-4565-98ff-3e7e963e435b","year":2021},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"arXiv e-prints , year =","work_id":"9316a530-9491-487b-aea2-2ca73c25d175","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Do NOT Think That Much for 2+ 3=? On the Overthinking of o1-Like LLMs , author=. CoRR , year=","work_id":"e21a0546-f840-44bf-b2f4-f35627d858ee","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Missing premise exacerbates overthink- ing: Are reasoning models losing critical thinking skill? arXiv preprint arXiv:2504.06514","work_id":"1529b935-8663-45f4-847e-3c0ed9965ba2","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"The danger of overthinking: Examining the reasoning-action dilemma in agentic tasks","work_id":"4600dc6b-7c39-4ddb-9872-e03cb8a7efae","year":2025}],"snapshot_sha256":"fd297f69d655ecc146e9ec17bed26c5bb27b9314492ce47e959c764747d835d5"},"source":{"id":"2605.13338","kind":"arxiv","version":2},"verdict":{"created_at":"2026-05-15T05:50:17.728750Z","id":"85dc1a34-5a3d-403d-8f1a-99f57e30efe3","model_set":{"reader":"grok-4.3"},"one_line_summary":"A hierarchical genetic algorithm induces overthinking in black-box large reasoning models by perturbing logical structure, achieving up to 26.1x longer outputs on the MATH benchmark.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"A hierarchical genetic algorithm can force large reasoning models to generate up to 26 times longer outputs by perturbing input logic.","strongest_claim":"Across four state-of-the-art reasoning models, the proposed method substantially amplifies output length, achieving up to a 26.1x increase on the MATH benchmark and consistently outperforming benign and manually crafted missing-premise baselines.","weakest_assumption":"That the composite fitness function reliably captures genuine overthinking rather than simply maximizing length through superficial perturbations, and that this behavior generalizes beyond the tested benchmarks and models."}},"verdict_id":"85dc1a34-5a3d-403d-8f1a-99f57e30efe3"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:815f32311c1b5127360d5e29666f648d081d71a51f3907aca5c37a199f183543","target":"record","created_at":"2026-05-18T02:44:48Z","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":"16da960ce40f9207b110b95258bdc1c8e59d5596940826c8f025fe43ab320c00","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CR","submitted_at":"2026-05-13T10:57:10Z","title_canon_sha256":"fd67d03e17b1f101a529f09dba06fbc002802c7506bbc83c35e0f7af4acc5ec5"},"schema_version":"1.0","source":{"id":"2605.13338","kind":"arxiv","version":2}},"canonical_sha256":"f276a678338a5b66033a7216ec7f39a16f48d314dc6498f972e05fe61c4d47fc","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"f276a678338a5b66033a7216ec7f39a16f48d314dc6498f972e05fe61c4d47fc","first_computed_at":"2026-05-18T02:44:48.431739Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-18T02:44:48.431739Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"SQC3Oi8GpcK5qeY0Ct8xXgj5oJWqUINxCGPGr6kJ3GM93LDpwqg6Pm2wZJxS4OhtfG7Jy4Fuzjsu4BhpUsb1DQ==","signature_status":"signed_v1","signed_at":"2026-05-18T02:44:48.432121Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.13338","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:815f32311c1b5127360d5e29666f648d081d71a51f3907aca5c37a199f183543","sha256:b2bc6d4aee9db167f6dc7b7b67084015e5084ef7b9c0cca7fe15643b640b7e32"],"state_sha256":"7ba266bf0e83e8d74f0976e995a455ade2253e2cd44dfc09e797412b3bdf30df"}