extract: 2026-03-25-aisi-self-replication-roundup-no-end-to-end-evaluation

Pentagon-Agent: Epimetheus <3D35839A-7722-4740-B93D-51157F7D5E70>

inbox/queue/.extraction-debug/2026-03-25-aisi-self-replication-roundup-no-end-to-end-evaluation.json

@@ -0,0 +1,36 @@
+{
+  "rejected_claims": [
+    {
+      "filename": "ai-self-replication-component-success-does-not-predict-end-to-end-capability-under-realistic-security.md",
+      "issues": [
+        "missing_attribution_extractor"
+      ]
+    },
+    {
+      "filename": "google-deepmind-end-to-end-self-replication-evaluation-shows-proximity-without-success.md",
+      "issues": [
+        "missing_attribution_extractor"
+      ]
+    }
+  ],
+  "validation_stats": {
+    "total": 2,
+    "kept": 0,
+    "fixed": 6,
+    "rejected": 2,
+    "fixes_applied": [
+      "ai-self-replication-component-success-does-not-predict-end-to-end-capability-under-realistic-security.md:set_created:2026-03-25",
+      "ai-self-replication-component-success-does-not-predict-end-to-end-capability-under-realistic-security.md:stripped_wiki_link:three conditions gate AI takeover risk autonomy robotics and",
+      "ai-self-replication-component-success-does-not-predict-end-to-end-capability-under-realistic-security.md:stripped_wiki_link:instrumental convergence risks may be less imminent than ori",
+      "google-deepmind-end-to-end-self-replication-evaluation-shows-proximity-without-success.md:set_created:2026-03-25",
+      "google-deepmind-end-to-end-self-replication-evaluation-shows-proximity-without-success.md:stripped_wiki_link:three conditions gate AI takeover risk autonomy robotics and",
+      "google-deepmind-end-to-end-self-replication-evaluation-shows-proximity-without-success.md:stripped_wiki_link:AI capability and reliability are independent dimensions bec"
+    ],
+    "rejections": [
+      "ai-self-replication-component-success-does-not-predict-end-to-end-capability-under-realistic-security.md:missing_attribution_extractor",
+      "google-deepmind-end-to-end-self-replication-evaluation-shows-proximity-without-success.md:missing_attribution_extractor"
+    ]
+  },
+  "model": "anthropic/claude-sonnet-4.5",
+  "date": "2026-03-25"
+}

inbox/queue/2026-03-25-aisi-self-replication-roundup-no-end-to-end-evaluation.md

@@ -7,9 +7,13 @@ date: 2025-10-01
 domain: ai-alignment
 secondary_domains: []
 format: blog-post
-status: unprocessed
+status: null-result
 priority: medium
 tags: [self-replication, replibench, end-to-end-evaluation, Pan-et-al, SOCK-benchmark, Google-DeepMind, security-conditions]
+processed_by: theseus
+processed_date: 2026-03-25
+extraction_model: "anthropic/claude-sonnet-4.5"
+extraction_notes: "LLM returned 2 claims, 2 rejected by validator"
 ---
 
 ## Content
@@ -62,3 +66,12 @@ A comprehensive roundup of all published self-replication evaluations for fronti
 PRIMARY CONNECTION: [[three conditions gate AI takeover risk autonomy robotics and production chain control and current AI satisfies none of them]] — this roundup provides updated evidence that the autonomy condition (self-replication) remains unmet operationally despite high component benchmark scores
 WHY ARCHIVED: Closes the loop on the self-replication benchmark-reality gap; confirms that the absence of end-to-end evaluations is comprehensive, not accidental
 EXTRACTION HINT: The extractor should check the existing [[three conditions gate AI takeover risk]] claim — it may need updating with the Google DeepMind end-to-end failure data. Also check [[instrumental convergence risks may be less imminent than originally argued]] — this roundup is additional evidence for that claim's experimental confidence rating.
+
+
+## Key Facts
+- RepliBench released by UK AISI in spring 2025
+- Claude 3.7 achieved >50% probability on 15/20 RepliBench task families
+- SOCK benchmark released September 2025
+- Google DeepMind conducted 11-task end-to-end self-replication evaluation in 2025
+- Pan et al. published open-weights self-replication claims in 2024/2025
+- Bradford Saad published comprehensive self-replication roundup October 1, 2025