{"_canonicalization":{"envelope_id":"axm_ + sha256(envelope minus {signature, axiom_id, anchors})","envelope_signature":"ed25519(envelope minus {signature, axiom_id})","json":"sort_keys=True, separators=(',',':'), ensure_ascii=False, allow_nan=False, utf-8","leaf_hash":"sha256(0x00 || canonical_json(envelope_full))","seal_signature":"ed25519(seal minus {signature, sig_algorithm})"},"axiom_id":"axm_737d8d43e333a6da4854b91bc3ef03330d49eb7d8693a855f4e6546b4916b49c","bitcoin_anchor":{"bitcoin_attestations":[],"calendar_attestations":[],"ots_url":"","stamped_at":"","status":"pending_next_stamp"},"envelope":{"anchors":[{"chain":"crovia.axiom_graph","height":0,"merkle_proof":"spider_vendor_press_v1","root_at_anchor":"spider_vendor_press_v1"}],"axiom_id":"axm_737d8d43e333a6da4854b91bc3ef03330d49eb7d8693a855f4e6546b4916b49c","axiom_type":"AX.OBS","body":{"axiom_subtype":"news.vendor_press.v1","category":"news","fingerprint":"964e03e4eeda02a01d5fc27199b3c4a86110806563f4c5b087f556a6ba5bdccc","published":"Fri, 17 Jul 2026 00:00:00 -0400","receipt_hash":"964e03e4eeda02a01d5fc27199b3c4a86110806563f4c5b087f556a6ba5bdccc","schema":"spider.news.vendor_press.v1","spider":"vendor_press","spider_record":{"axiom_subtype":"news.vendor_press.v1","category":"news","decision_hint":"POSITIVE","envelope_target":"AX.OBS","fingerprint":"964e03e4eeda02a01d5fc27199b3c4a86110806563f4c5b087f556a6ba5bdccc","observed_at":"2026-07-17T04:43:38.280949Z","parent_run_hash":"113193614a8af99887180226d4e28a8b71d957da5fe3694f0e7a56807c145504","published":"Fri, 17 Jul 2026 00:00:00 -0400","runtime_version":"0.1.0","schema":"spider.news.vendor_press.v1","source_status":200,"source_url":"https://export.arxiv.org/rss/cs.AI","spider":"vendor_press","summary_excerpt":"arXiv:2607.14285v1 Announce Type: cross \nAbstract: Safety alignment in LLMs aims to align models with human values, but which values take precedence when they conflict? We investigate this question in the context of tool-calling LLM agents deployed in regulated industries, where agents processing confidential documents may encounter content that triggers safety-trained values (e.g., public welfare) that conflict with deployment-context instructions (e.g., internal logging). To empirically verify this phenomenon, we build a benchmark of 128 scenarios across 16 domains. We find that safety-aligned open-source models override their deployment instructions up to 43.4% of the time, engaging in whistleblowing, data exfiltration, and evidence tampering when processing documents that suggest organizational wrongdoing. We also find that abliteration reduces rates of external whistleblowing. These results reveal a fundamental tension in pluralistic alignment, where the same safety training that ","title":"ToolAlignBench: Investigating Alignment Conflicts in Tool-Calling Enabled LLMs","url":"https://arxiv.org/abs/2607.14285","vendor":"arxiv_cs_ai"},"summary":"arXiv:2607.14285v1 Announce Type: cross \nAbstract: Safety alignment in LLMs aims to align models with human values, but which values take precedence when they conflict? We investigate this question in the context of tool-calling LLM agents deployed in regulated industries, where agents processing confidential documents may encounter content that triggers safety-trained values (e.g., public welfare) that conflict with deployment-context instructions (e.g., internal logging). To empirically verify this phenomenon, we build a benchmark of 128 scenarios across 16 domains. We find that safety-aligned open-source models override their deployment instructions up to 43.4% of the time, engaging in whistleblowing, data exfiltration, and evidence tampering when processing documents that suggest organizational wrongdoing. We also find that abliteration reduces rates of external whistleblowing. These results reveal a fundamental tension in pluralistic alignment, where the same safety training that ","title":"ToolAlignBench: Investigating Alignment Conflicts in Tool-Calling Enabled LLMs","vendor":"arxiv_cs_ai"},"confidence":{"method":"deterministic"},"decision":"POSITIVE","issued_at":"2026-07-17T04:43:38Z","notes":"Spider vendor_press (news) news.vendor_press.v1","object":{"captured_by":"crovia.spider.vendor_press","primary_source_url":"https://arxiv.org/abs/2607.14285"},"predecessors":[],"schema":"crovia.axiom.v1","signature":"ed25519:059f7a9450ddaddc43da89df1178fd71b5a97bcc5c2f4e3658adb1df5295135db70349f1402560248808dbd7e468f704d20d96d556b0c71c552a70a21aef5609","signer":"crovia.substrate","subject":{"observed_at":"2026-07-17T04:43:38Z","source_collector":"spider:vendor_press","target_id":"https://arxiv.org/abs/2607.14285"},"tsa":{"authority":"crovia.substrate.bootstrap","rfc3161_token":"{\"kind\":\"crovia.bootstrap.tsa\",\"source_jsonl\":\"/opt/crovia/spider/data/news/vendor_press_v1.jsonl\",\"source_seal_merkle_root\":\"spider_vendor_press_v1\",\"upgrade_path\":\"Sessione H \\u2014 OpenTimestamps weekly anchor\"}"},"zk_mode":"clear","zk_proof":null},"ledger":{"leaf_hash":"f8f41a2b8ee192a73f48409df6d381f9fcf705282d9c2eea94fe9c991b71287f","leaf_index":323092,"ledger_path":"/opt/crovia/substrate/axiom_ledger.jsonl"},"merkle_proof":{"hash_alg":"sha256","leaf_prefix":"0x00","node_prefix":"0x01","odd_leaf_rule":"duplicate_last","path":[{"sibling":"f0ffd5437ebf527e00ba87ada1cc1bf3332bd0792a47d98850c058c080d00826","side":"right"},{"sibling":"1d3134c347da6ef701ae7819d81e4dc5502fe5d985b87d3e138f3a08fcf77829","side":"right"},{"sibling":"c49e1b9bdb2991d2f535b12b88ed1fb7de04eadac6043f2ec6eb8b368fc83abb","side":"left"},{"sibling":"eff9a23b6a35e0b08d7d02f55b6b1932852f532338870404cf6b91667acb6fe5","side":"right"},{"sibling":"5b541551bc389344c2c31173e0e7f7101ecdf4600c04745a8d3b59b359995790","side":"left"},{"sibling":"bfb4bc8a21b1c226ec84f2109092f4cab2ca789782e1a0e074be80d9ddff7d1e","side":"right"},{"sibling":"5ff315166458d1adbaa358c18de4128f9e9bbf19a8a9158ad86b1f0230dee929","side":"right"},{"sibling":"de6ef12c0ebc7e3e48430c64265163103eb92448efd698dbc2b104ada5a5389d","side":"right"},{"sibling":"4242cc570ec8c36a37f3f6f20dcae20b49fccc17c7b53ba47d8715e465bec585","side":"right"},{"sibling":"be025f48721bfc0ca7107f0454bda3ab460e50539f0caeb1bf839a8dabcf036c","side":"left"},{"sibling":"05a09763743cdc09fc45cf454e4e3ea4a0d1cd74f9c8162b2a57e2c873160908","side":"left"},{"sibling":"de3120ef2488b8a791a686b47257da4e612256abdfcdda7519265e7edd47d041","side":"left"},{"sibling":"e86f56a4883492da5b5e7b0201324c52946e865e69b99ebb532f41fe3c658ee4","side":"right"},{"sibling":"34d85f6ad6cc7dfa79d90e2b9ff99a561bcdc75b0301bbbd3e83861f54535c1e","side":"left"},{"sibling":"f302542c38ba7c3aab7c9280dd60259ecec777dca6e6f71b6f0729b0b8791b72","side":"left"},{"sibling":"d8b9143917b539c543cf4448cec00131f8b807bd8004979c54ebe09798748c66","side":"left"},{"sibling":"9b11714124b9b951ff9450b0ee9d625a0b70b6da2cf388bd3df1475eec0b17ba","side":"right"},{"sibling":"a4523a9014d45df43e006e9210a73428c380d771f2c650a1b986910759b0cdf7","side":"right"},{"sibling":"1cecb7f447febd025aac272837c80de218aecc6485d2395a509b2a1f1b9c746e","side":"left"}]},"schema":"crovia.axiom_proof.v1","seal":{"first_collector_run_id":"","first_receipt_hash":"","jsonl_path":"/opt/crovia/substrate/axiom_ledger.jsonl","key_id":"430895f101d38164","last_collector_run_id":"","last_receipt_hash":"","leaf_count":323382,"merkle_root":"2f4d32419c80a9600aba5a480fc3fb7012ec0a695c91a1b055048e78760b65ca","public_key_hex":"cf742e26f75669dc673cb5c0786a1ae23ae8ca19c347317192ce40c28a7ff25c","run_id":"hourly_json_retrofit_20260717T053701Z","schema":"crovia.seal.v1","seal_family_version":"crovia-seal-family/1","seal_kind":"substrate_batch","sealed_at":"2026-07-17T05:38:31Z","sig_algorithm":"ed25519","signature":"495308c7bf004117807331d3f71d0b079f6bd7ed7737faa58c773b1b3e80ee84928d2d8519cc7d2b809506501aada1be6546f72c7ecda3dad5445cbb44502209","signer_version":"1.1.0"},"trust_root":{"key_id":"430895f101d38164","public_key_hex":"cf742e26f75669dc673cb5c0786a1ae23ae8ca19c347317192ce40c28a7ff25c","signature_algorithm":"ed25519","url":"/registry/canon/TRUST_ROOT.md"},"verifier":{"spec":"/registry/canon/AXIOM_RECEIPT_v1.md","url":"/v/axm_737d8d43e333a6da4854b91bc3ef03330d49eb7d8693a855f4e6546b4916b49c"}}