{"_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_09acd5bac536a0fe7c78c747d341a330492c518cbc9f3e567361f5574bdf4336","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_09acd5bac536a0fe7c78c747d341a330492c518cbc9f3e567361f5574bdf4336","axiom_type":"AX.OBS","body":{"axiom_subtype":"news.vendor_press.v1","category":"news","fingerprint":"74f043961f132cd27240b4e8160189d2c8876b68957d8d7fc556a6b24ac433fc","published":"Fri, 03 Jul 2026 00:00:00 -0400","receipt_hash":"74f043961f132cd27240b4e8160189d2c8876b68957d8d7fc556a6b24ac433fc","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":"74f043961f132cd27240b4e8160189d2c8876b68957d8d7fc556a6b24ac433fc","observed_at":"2026-07-03T04:43:38.241623Z","parent_run_hash":"f0e30469786257a5e74170498cacb4c028623bf32d6d06d4dbadc488960545be","published":"Fri, 03 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:2607.02020v1 Announce Type: new \nAbstract: Multimodal large language models must continually adapt to evolving tasks and domains, yet standard continual learning metrics mainly measure whether old answers remain correct, leaving the stability of multimodal grounding largely unexamined. We study this overlooked failure mode and ask whether a continually adapted MLLM can preserve not only what it answers, but also how it uses visual, textual, OCR, chart, and document evidence. We identify \\emph{hidden evidence-use forgetting}, where answer accuracy is retained while the model silently shifts toward different or less grounded evidence channels, and propose \\textsc{RCL}, a replay-free reliance-constrained continual learning framework. \\textsc{RCL} freezes the previous checkpoint as a behavioral reference, estimates teacher and student evidence-reliance profiles through counterfactual channel interventions, and jointly optimizes task learning, prediction preservation, and reliance pre","title":"Hidden Forgetting in Continual Multimodal Learning: When Accuracy Survives but Grounding Fails","url":"https://arxiv.org/abs/2607.02020","vendor":"arxiv_cs_ai"},"summary":"arXiv:2607.02020v1 Announce Type: new \nAbstract: Multimodal large language models must continually adapt to evolving tasks and domains, yet standard continual learning metrics mainly measure whether old answers remain correct, leaving the stability of multimodal grounding largely unexamined. We study this overlooked failure mode and ask whether a continually adapted MLLM can preserve not only what it answers, but also how it uses visual, textual, OCR, chart, and document evidence. We identify \\emph{hidden evidence-use forgetting}, where answer accuracy is retained while the model silently shifts toward different or less grounded evidence channels, and propose \\textsc{RCL}, a replay-free reliance-constrained continual learning framework. \\textsc{RCL} freezes the previous checkpoint as a behavioral reference, estimates teacher and student evidence-reliance profiles through counterfactual channel interventions, and jointly optimizes task learning, prediction preservation, and reliance pre","title":"Hidden Forgetting in Continual Multimodal Learning: When Accuracy Survives but Grounding Fails","vendor":"arxiv_cs_ai"},"confidence":{"method":"deterministic"},"decision":"POSITIVE","issued_at":"2026-07-03T04:43:38Z","notes":"Spider vendor_press (news) news.vendor_press.v1","object":{"captured_by":"crovia.spider.vendor_press","primary_source_url":"https://arxiv.org/abs/2607.02020"},"predecessors":[],"schema":"crovia.axiom.v1","signature":"ed25519:ee50cfcc239295da421d9569695ceb2464437ceac079b2914568c2810acb8bd8158ed716d4d251541bd2b90e1752dd3021eb6b6602fe823c8a764070e88d5205","signer":"crovia.substrate","subject":{"observed_at":"2026-07-03T04:43:38Z","source_collector":"spider:vendor_press","target_id":"https://arxiv.org/abs/2607.02020"},"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":"98f51e626b86c9ba6c0b9b71d0b6a8f94776f323d612d84083bdef1d4965604e","leaf_index":275392,"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":"540ffa73dda59d692a41bcae998aaf2e6495ef177dff7c25c4333e89086bbd71","side":"right"},{"sibling":"bcc1b35cf3d9e43a3b49d9ee631b74d4d31424317bbb3dca81b6e804e519b13b","side":"right"},{"sibling":"abba8d450331da29936821bc1184a0f4ad74c2143aabe79b848bb135c66cb44a","side":"right"},{"sibling":"1da9d236c72cc5e1121662636d9c0a7ce46d55227287a77f4b9440d887af08dc","side":"right"},{"sibling":"5d8a5029c968d0f123b5bd1701855306d2cd60dd310556cfb484e937c3d1cde5","side":"right"},{"sibling":"849f8f2e2bcf09437f8fb74995be42a57cbd6b5bbf3b4d1b680700e05d9f3916","side":"right"},{"sibling":"3a4dd616b0a708226bdfc1375ba2f51d5b9f5745b3184702e9cbca38e27d2a27","side":"left"},{"sibling":"fd055e725b36f5bebbaeb18583f9d7011c4d8c2a608a11cbd2c166cd895e7a85","side":"left"},{"sibling":"e1a65581e9d68211aa8ba0d138c8d2a4e80afd551e6d1b00c4575c39085df705","side":"left"},{"sibling":"92c062f377cd53076b8dfe55c25ab217059e20113433673cc5747b9348b07e01","side":"left"},{"sibling":"e36f7633c67452f41a7377a7bec9b2d454442688d26539321eebf92cae9db0f0","side":"right"},{"sibling":"4dbd8247ba08a5432c7d6540711da9acb2f59e6189865aa8552dee37f69286a9","side":"right"},{"sibling":"41cd1885dc3fcb51e49eeb887d6d22ec2cfa58df0e4f8d7c7dddf3a1b0ce8249","side":"left"},{"sibling":"8a09562f6b247c1c3cd1fea36cb3b8f1cf5c575479dd514573856a380a964bf5","side":"left"},{"sibling":"723981908169653ca6d835aa9b8381a8c7ad3e3e3830d0792bc32032cda615ee","side":"right"},{"sibling":"c0594fa1ee81d5f019cccc7b5e51af603c6d7e43995498c451012060c7d06165","side":"right"},{"sibling":"4de6a2fb22efbb50c84dc62abeb0f2cbc8c663a9540aeba9e758ebfdfe3e86dd","side":"right"},{"sibling":"fdbb3519f8dc411a4043dfb5abdbfea5441e130326183ac2247c42584033f152","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":275799,"merkle_root":"2581d0d6e5fa345cdf2e8ab3b191ace76d6b14189901ab0e4c2291ca1d1ae1e6","public_key_hex":"cf742e26f75669dc673cb5c0786a1ae23ae8ca19c347317192ce40c28a7ff25c","run_id":"hourly_json_retrofit_20260703T053701Z","schema":"crovia.seal.v1","seal_family_version":"crovia-seal-family/1","seal_kind":"substrate_batch","sealed_at":"2026-07-03T05:38:09Z","sig_algorithm":"ed25519","signature":"e44a386a5ae00c0e7fc67b1179bb9060bf0fefc006e454fec70e27668182ff497d1e2faf0c3b22de0917beefb7c80e880dae925a3f67e1b16aa0eb44bf947407","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_09acd5bac536a0fe7c78c747d341a330492c518cbc9f3e567361f5574bdf4336"}}