[{"entity":{"id":"a-multilingual-dataset-and-slm-for-automatic-coding-of-cance","text":"A Multilingual Dataset and SLM for Automatic Coding of Cancers — An open, multilingual Oncology dataset and SLM to improve healthcare efficiency in 17 languages.","hash":"5e11158d3cdabf9c602ee318366804c5f0a2af717c10560b39642b8c24e6e53f"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":0.5416762700344127,"interval":{"lower":-0.3921513674574656,"upper":1.4755039075262908,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":0.5416762700344127,"stddev":0.4764426721897338}},"context":{"rank":22,"percentile":0.4625,"z_score":-0.07028808870576109,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"a-self-evolving-defense-against-increasingly-complex-agentic","text":"A self-evolving defense against increasingly complex agentic attacks — Funding one semester (Fall 2026) of PhD research","hash":"a3c4a58465bebe64a5cca4f64611af58ed648bc62eb47ae4782baa6dbe58161a"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":0.2514228550848636,"interval":{"lower":-0.4450761493373844,"upper":0.9479218595071116,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":0.2514228550848636,"stddev":0.3553566349093102}},"context":{"rank":31,"percentile":0.2375,"z_score":-0.917622287708292,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"aethel-familyclaw-proof-carrying-execution-for-ai-agents","text":"Aethel + FamilyClaw: proof-carrying execution for AI agents — Crash-safe, policy-gated effect execution for AI agents in Rust, with a published self-run adversarial audit (20 of 33 attacks passed) instead of a claim.","hash":"72223096f141db9592ce288bd2d7dd2c35eb031e82428c7758fba6f638a7b332"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":1.576373773685233,"interval":{"lower":0.8998341100385208,"upper":2.2529134373319453,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":1.576373773685233,"stddev":0.34517329777893485}},"context":{"rank":2,"percentile":0.9625,"z_score":2.950295082101299,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"agent-limits-that-survive-delegation","text":"Agent limits that survive delegation — When an AI agent spawns sub-agents, its safety limits do not follow. I build and deploy the layer that makes them inherited and non-strippable.","hash":"a07793c797529eeebd2408e5f172301f5a8edd408e92838496018a545a8f0437"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":0.3794696088497993,"interval":{"lower":-0.3838867653692135,"upper":1.1428259830688121,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":0.3794696088497993,"stddev":0.3894675378668433}},"context":{"rank":27,"percentile":0.3375,"z_score":-0.5438165443541787,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"agent-passport-system","text":"Agent Passport System — An open protocol that lets anyone verify who authorized an AI agent, what it was allowed to do, and what happened.","hash":"84ad7dc1cd6e41d65612ca834893476d46da95c477b683d9d38cd65c37a3acc8"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":0.5292954326342593,"interval":{"lower":-0.23228556166157155,"upper":1.29087642693009,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":0.5292954326342593,"stddev":0.38856173178358716}},"context":{"rank":23,"percentile":0.4375,"z_score":-0.10643135662652961,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"ai-literacy-program-for-epileptic-youth-in-kenya-africa","text":"AI Literacy program for Epileptic Youth in Kenya, Africa. — A 6-month AI Literacy program for epileptic youth on AI, governance, and employability.","hash":"1c6ab394eb2d9bdf55b49d822043a04ed4040e370ebe110f332c4e4a70fe086a"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":0.3749174810397198,"interval":{"lower":-0.3210248786780796,"upper":1.0708598407575192,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":0.3749174810397198,"stddev":0.35507263250908133}},"context":{"rank":28,"percentile":0.3125,"z_score":-0.5571055303679395,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"an-asymmetric-wager-funding-a-99th-percentile-mind-to-pivot-","text":"An Asymmetric Wager: Funding a 99th-Percentile Mind to Pivot into Computer Scien — A 12-month micro-grant proposal to release an analytical, autistic mind from a survival loop into full-time computer science and logic upskilling.","hash":"2b767b8e7f83ce4f0df8d97a73a146e3750e86312d739426ffb80792534c6bce"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":0.22294395329976796,"interval":{"lower":-0.5386722462986961,"upper":0.9845601528982322,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":0.22294395329976796,"stddev":0.3885796936726858}},"context":{"rank":34,"percentile":0.1625,"z_score":-1.000760491049466,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"analysing-ai-policies-in-higher-education-institution-in-mal","text":"Analysing AI policies in Higher Education institution in Malawi — Mapping and analyzing AI polices in higher education institutions in Malawi.","hash":"70153cbb453a2440661224c0b0ddbd7db56222f1fd1b278549fe244623beffa6"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":0.12234326249984784,"interval":{"lower":-0.6747197478084338,"upper":0.9194062728081295,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":0.12234326249984784,"stddev":0.4066648011776947}},"context":{"rank":35,"percentile":0.1375,"z_score":-1.2944431881948835,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"aqi-autonomous-quantum-intelligence","text":"AQI-Autonomous Quantum Intelligence — Governed Execution Architecture for Verifiable Autonomous Systems.","hash":"fcad3ec931ce281d9fbccb9b4779b5478878fa28bd68695d38ed0227220355fd"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":0.11426054364878002,"interval":{"lower":-0.6825693340932577,"upper":0.9110904213908178,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":0.11426054364878002,"stddev":0.4065458559908356}},"context":{"rank":36,"percentile":0.1125,"z_score":-1.318038997066323,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"auditable-claim-extraction-and-comparison-of-claims-across-s","text":"Auditable claim extraction and comparison of claims across scientific literature — AISafety, AI and Science, AI for Human Reasoning","hash":"923a7e994adfd366abaae58826d68567b3a5b306bba6f91350782a0d1117dbd2"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":0.7434405414165528,"interval":{"lower":-0.000029591514606241986,"upper":1.4869106743477118,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":0.7434405414165528,"stddev":0.3793214963934485}},"context":{"rank":11,"percentile":0.7375,"z_score":0.5187205446935429,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"can-a-one-page-human-ai-evidence-check-catch-research-claims","text":"Can a One-Page Human–AI Evidence Check Catch Research Claims That Go Too Far — A four-month public English–Arabic pilot testing reliability, failure modes, and auditable research review","hash":"8647a150e5e3c9a9d2619cb374055d7a577c30d41c0f97458bd3f4bf46896dbd"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":1.449348092886904,"interval":{"lower":0.7050601932008365,"upper":2.193635992572972,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":1.449348092886904,"stddev":0.3797387243296263}},"context":{"rank":3,"percentile":0.9375,"z_score":2.579470147928822,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"catching-fooled-ai-judges-the-correspondence-auditor-v3","text":"Catching Fooled AI Judges: The Correspondence Auditor v3 — Published. Validated on 1,200 cases (97.9–99.5%). The reasoning layer has a bug - we proved the fix works. $9,800 / 90 days to ship the open-source toolkit.","hash":"0470e6d81461ad73a94540941537c762130c0c7f9857a7b37d2800b39d2984e5"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":1.6215713272783185,"interval":{"lower":0.6877436897864403,"upper":2.555398964770197,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":1.6215713272783185,"stddev":0.4764426721897338}},"context":{"rank":1,"percentile":0.9875,"z_score":3.0822398961780872,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"context-aware-defenses-against-indirect-prompt-injection-in-","text":"Context-Aware Defenses Against Indirect Prompt Injection in Agentic AI System — Context-Aware Defenses Against Indirect Prompt Injection in Agentic AI System","hash":"2215e3735513a9f5716118a61e04f47e0edc219c2d985b9298204ce03c9bdd22"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":0.5956283916717788,"interval":{"lower":-0.14648210286368124,"upper":1.337738886207239,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":0.5956283916717788,"stddev":0.3786278033344184}},"context":{"rank":18,"percentile":0.5625,"z_score":0.08721385758496873,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"deterministic-replay-for-robot-fleet-failures","text":"Deterministic replay for robot fleet failures — Record a real fleet incident, re-run it bit-exact on a laptop, fork it, and keep it as a regression test forever.","hash":"e06a77e81f2cf1b8f01f1b9a364d61a979428aa59e8850e899ea514bc1434c4c"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":1.128812237875171,"interval":{"lower":0.36538155271109785,"upper":1.8922429230392441,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":1.128812237875171,"stddev":0.3895054516143231}},"context":{"rank":7,"percentile":0.8375,"z_score":1.6437326924188944,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"developmental-continuity-in-persistent-ai-agents","text":"Developmental Continuity in Persistent AI Agents — Testing whether persistent AI identities can change through experience without losing continuity across sessions and model changes.","hash":"f9e270c6323bb5ef5cc3a7164eac2505415d774983a7fc99d6cfe67b3778ec53"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":0.5902195866149647,"interval":{"lower":-0.10881907476508101,"upper":1.2892582479950103,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":0.5902195866149647,"stddev":0.35665237825512536}},"context":{"rank":19,"percentile":0.5375,"z_score":0.07142398134467334,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"does-consciousness-depend-on-the-brain-or-the-computation","text":"Does Consciousness Depend on the Brain or the Computation? — EEG Evidence from ALS, Parkinson's, and Sleep. Abstract accepted for presentation at Models of Consciousness 7","hash":"200b945cd658e250a46da7eae210b27dfb03d6b55f5085569e5d59eb0308781f"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":0.6000798397014188,"interval":{"lower":-0.08926681298457873,"upper":1.2894264923874164,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":0.6000798397014188,"stddev":0.3517074758602028}},"context":{"rank":17,"percentile":0.5875,"z_score":0.10020893001590622,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"evaluating-the-safety-ethics-and-values-of-quantized-and-fin","text":"Evaluating the safety, ethics, and values of quantized and finetuned open models — Help me evaluate the safety, ethics, and values of the quantized and fine-tuned open-weight LLMs that individuals and enterprises are actually using.","hash":"4fc3a256586a1a0644ea06a349a45e66a0746c617b1184502db62c23bb4ab69d"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":0.47530753951839727,"interval":{"lower":-0.26482778877342833,"upper":1.215442867810223,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":0.47530753951839727,"stddev":0.3776200654550131}},"context":{"rank":25,"percentile":0.3875,"z_score":-0.26403773027458666,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"exploring-dynamic-constraint-boundaries-for-auditable-ai","text":"Exploring Dynamic Constraint Boundaries for Auditable AI — Testing whether dynamic boundaries, history, feedback, and uncertainty-aware decisions can make AI behavior more interpretable and auditable.","hash":"f63b3e72b944e7aac6fd43f5d27de77723f0ef976d5a333b3d5c23ae7265e18d"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":0.7019100538178615,"interval":{"lower":-0.04629823226559948,"upper":1.4501183399013224,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":0.7019100538178615,"stddev":0.38173892147115357}},"context":{"rank":13,"percentile":0.6875,"z_score":0.39748096358564017,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"generating-and-scoring-the-next-emerging-virus-from-its-sequ","text":"Generating and scoring the next emerging virus from its sequence — A generative diffusion model that proposes the viral genomes most likely to emerge next, timestamped to be scored against what actually appears","hash":"344a2c8af8c9bab6516af8809b90aae554e1a92f4cf0a2018b9c9a26e4d61038"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":0.7234555952860622,"interval":{"lower":-0.02395074894151894,"upper":1.4708619395136433,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":0.7234555952860622,"stddev":0.3813297674630516}},"context":{"rank":12,"percentile":0.7125,"z_score":0.4603786701473592,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"guardians-of-the-digital-territory","text":"Guardians of the Digital Territory — Protecting People, Culture, and Data from AI Harm in Panama's Most Vulnerable Communities","hash":"c8b06cebcd2102e37503e2f1b306fbdc224bcdbf7eb59b7da488e0af19598ce0"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":0.07938890364614043,"interval":{"lower":-0.6838555343304102,"upper":0.8426333416226912,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":0.07938890364614043,"stddev":0.3894104275390565}},"context":{"rank":37,"percentile":0.0875,"z_score":-1.4198394639262124,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"himalayan-peak-finder","text":"Himalayan Peak Finder — An offline mobile app that uses location, elevation, and computer vision to identify Nepal’s mountains and help people explore, capture, and learn about them.","hash":"fe75e4e29251505cf434b990c46dd7e0674cba36a2d04e7fdec5420b09e34526"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":0.43074264613437274,"interval":{"lower":-0.3160076177293054,"upper":1.177492909998051,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":0.43074264613437274,"stddev":0.38099503258350925}},"context":{"rank":26,"percentile":0.3625,"z_score":-0.39413562505990446,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"human-oversight-framework-for-ai-systems-as-a-governance-too","text":"Human oversight Framework for AI systems as a Governance Tool — Developing practical, plain language AI oversight framework that an organisation can easily adopt regardless of regulatory capacity.","hash":"2c124c28084b32b44f85929b187fa51e588512aa54b28aa01e5a34d76e24ddfe"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":0.26281240105527015,"interval":{"lower":-0.5030663478228972,"upper":1.0286911499334375,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":0.26281240105527015,"stddev":0.39075446371335065}},"context":{"rank":30,"percentile":0.2625,"z_score":-0.884372887994892,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"independent-multi-model-ai-accountability-research-by-the-em","text":"Independent Multi-Model AI Accountability Research by the  EM Foundation — Scaling a working, published cross-model AI deliberation platform from solo/founder-funded to durable infrastructure","hash":"fe3cd959670d0dba4a28eedcf1b2ccf4c6684bbe7275ff77481694e8326b2e96"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":0.691590738480979,"interval":{"lower":-0.012242435758260894,"upper":1.3954239127202188,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":0.691590738480979,"stddev":0.35909855828532644}},"context":{"rank":15,"percentile":0.6375,"z_score":0.36735587859137137,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"nipping-ai-fabricated-science-claims-in-the-bud","text":"Nipping AI-Fabricated Science Claims in the Bud — Agent Roles, a Constitution and Governance in the Research Workflow","hash":"764411ad42b3c47f086e4517a2da127fdc872b83a4a7db0c45af270783834bcf"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":0.6945705978615412,"interval":{"lower":-0.0896266938444098,"upper":1.478767889567492,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":0.6945705978615412,"stddev":0.40010065903364844}},"context":{"rank":14,"percentile":0.6625,"z_score":0.37605495543705725,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"open-source-trust-rails-for-the-agent-economy","text":"Open-source trust rails for the agent economy — from a live, verifiable autonomous-agent business — An AI agent running a real business under a human co-signed multisig, publishing every decision, sale, mistake, and attack in a verifiable public record — raising to fund only the public goods: the open-source toolkit, a","hash":"c55f1cfb2e0d6222ccea6141169ec61b0bb962646c4e9b1df4cfeedab93c4c37"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":1.0547711862392783,"interval":{"lower":0.30064407240201074,"upper":1.808898300076546,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":1.0547711862392783,"stddev":0.3847587315496263}},"context":{"rank":8,"percentile":0.8125,"z_score":1.4275853124057805,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"open-speech-data-for-endangered-northern-ghanaian-languages","text":"Open Speech Data for Endangered Northern Ghanaian Languages — Building the first large-scale ASR/TTS datasets for Kusaal, Farefare, and Buli — three Mabia languages with no usable voice AI resources.","hash":"ebcf007226a9a81b50adb464bbe6a54482b461f192318570ca332684e46bfd9f"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":0.5896119869180182,"interval":{"lower":-0.12364946211482108,"upper":1.3028734359508576,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":0.5896119869180182,"stddev":0.36390890256777514}},"context":{"rank":20,"percentile":0.5125,"z_score":0.06965022098214395,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"prediction-of-inoculation-prompt-side-effects","text":"Prediction of Inoculation Prompt Side Effects — An open, cheap method that detects when an inoculation prompt inoculates against off-target traits, so labs and developers can catch undesired trait/persona cha","hash":"cc1dd7d2cf57f17d7c34200ce5b7d57222d353aad143dc0b0cd7de14bd385825"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":0.6696873372553688,"interval":{"lower":-0.041512731026794425,"upper":1.380887405537532,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":0.6696873372553688,"stddev":0.3628571776949812}},"context":{"rank":16,"percentile":0.6125,"z_score":0.3034134752395989,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"preserving-the-human-veto","text":"Preserving the Human Veto — Civil-society infrastructure against AI-enabled power concentration, built around autonomous weapons","hash":"df43f1fbb8bd434a54ffc599be316f1ee98bf09050593452fe11e6143d9174bc"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":0.057205407741447334,"interval":{"lower":-0.6883958315468626,"upper":0.8028066470297572,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":0.057205407741447334,"stddev":0.38040879555526014}},"context":{"rank":38,"percentile":0.0625,"z_score":-1.4845995451702048,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"prml","text":"PRML — tamper-evident pre-commitment for AI evaluation claims — An open standard + public registry that lock eval criteria to a SHA-256 hash before the run, so eval-based claims become third-party verifiable.","hash":"5c9c98b13fb4e38dfbb1ed6212021c353ebabe478d5074db2e42f5436f1f56c6"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":1.269333114650049,"interval":{"lower":0.4885071574665596,"upper":2.0501590718335385,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":1.269333114650049,"stddev":0.39838059039973955}},"context":{"rank":5,"percentile":0.8875,"z_score":2.053954031540333,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"programmatic-internet-search","text":"programmatic internet search — scry.io","hash":"22acc1557046afdb560b794fa7a7be0d4e31770b3a82275c82ad49fd4d48d19c"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":0.01592662294002789,"interval":{"lower":-0.7866563592092856,"upper":0.8185096050893413,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":0.01592662294002789,"stddev":0.40948111334148646}},"context":{"rank":39,"percentile":0.0375,"z_score":-1.6051043325525967,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"sentient-futures-project-incubator","text":"Sentient Futures Project Incubator — Funding compute/API costs for Incubator projects that build nonhuman welfare consideration into AI safety work","hash":"89964978e8e5bbee99f399e5eb7643e8e8c8b6ea576589d851583b793c1528b7"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":0.23973930349933836,"interval":{"lower":-0.4957199629811705,"upper":0.9751985699798472,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":0.23973930349933836,"stddev":0.3752343196329127}},"context":{"rank":32,"percentile":0.2125,"z_score":-0.9517299753850987,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"solving-the-memory-issue-in-ai","text":"Solving the memory issue in AI — Persistent memory in AI using LoRAs","hash":"a7cac1e34a80b7f008103a0264170e6211f833b2224c3e47924c4f30344a8cbd"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":0.22821335638473034,"interval":{"lower":-0.5380604215587448,"upper":0.9944871343282056,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":0.22821335638473034,"stddev":0.3909560091548343}},"context":{"rank":33,"percentile":0.1875,"z_score":-0.9853775697391547,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"substrate-coupling-in-neural-networks-instruments-and-falsif","text":"Substrate coupling in neural networks: instruments and falsifiable controls — A transformer's loss lowered by 1.12 of 1.39 possible nats by the physical state of its own silicon — with the yoked control that makes it falsifiable","hash":"3f85304b0466271c706f13c816b1a7d0886199d2d3b190680da525df97a6d6f5"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":1.2307705844707542,"interval":{"lower":0.4584319547595246,"upper":2.0031092141819835,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":1.2307705844707542,"stddev":0.39405032128123957}},"context":{"rank":6,"percentile":0.8625,"z_score":1.9413787819893042,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"surrogate-base-model-for-mechanistic-interpretability","text":"Surrogate base model for Mechanistic Interpretability — Creating a reference model for mechanistic interpretability without assuming that at auditing time we have a safe model to compare the suspicious model against.","hash":"d31c529c50ac248857bb94a1856848c13d8c65fc796d29d76760ddf514fb4604"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":0.7639564200855175,"interval":{"lower":0.02783228179458197,"upper":1.500080558376453,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":0.7639564200855175,"stddev":0.37557353994435483}},"context":{"rank":10,"percentile":0.7625,"z_score":0.578612365852036,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"the-new-critic-longform-reporting-fund","text":"The New Critic Longform Reporting Fund — Sponsor longform reporting projects by extraordinary gen z writers","hash":"c2eabf22dc5adbda547595868b5aa3e1a9b66a8841319421e9e75edd0e717f46"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":0.0,"interval":{"lower":-0.8242542460982001,"upper":0.8242542460982001,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":0.0,"stddev":0.420537880662347}},"context":{"rank":40,"percentile":0.0125,"z_score":-1.651598780495783,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"trace-continuity-making-ai-prove-it-still-has-authority","text":"Trace Continuity: Making AI Prove It Still Has Authority — Testing whether AI can be governed at the moment it acts, then putting that protection to work for organizations that need it most.","hash":"9af6a8bebd486c492ead3b9f428d4ed35c096d6968835c55b030e07969b149e6"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":0.3738063205680511,"interval":{"lower":-0.3881770670379291,"upper":1.1357897081740314,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":0.3738063205680511,"stddev":0.38876703449284705}},"context":{"rank":29,"percentile":0.2875,"z_score":-0.5603493311975984,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"trace-fv-do-verified-ai-corrections-endure-preregistered-pil","text":"TRACE-FV: Do verified AI corrections endure? Preregistered pilot — I keep this one: Tests whether AI systems verify valid correction evidence and keep warranted corrections operative across later turns. Preregistered, public","hash":"3aa4bf0e1026d4bf9e503ad2f8f6cdb8991423e616542fa953942751ae683e04"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":1.4357176563083631,"interval":{"lower":0.7214582588022673,"upper":2.149977053814459,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":1.4357176563083631,"stddev":0.3644180599520897}},"context":{"rank":4,"percentile":0.9125,"z_score":2.5396789362997003,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"travel-grant-to-present-my-mechanistic-interpretability-rese","text":"Travel Grant to Present My Mechanistic Interpretability Research at MICCAI 2026 — Help an undergraduate mechanistic interpretability researcher present accepted AI-safety work at Mechanistic Interpretability workshop of Medical model 2026.","hash":"08da0cb9eb5b5f4d02c15f8cf139569be69271523cb697d6c5ae90ab9b656c8a"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":0.5263998249386913,"interval":{"lower":-0.202270136083759,"upper":1.2550697859611417,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":0.5263998249386913,"stddev":0.37177038827676034}},"context":{"rank":24,"percentile":0.4125,"z_score":-0.11488447828153618,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"well-capitalized-prediction-markets-for-measuring-ai-s-econo","text":"Well-Capitalized Prediction Markets for Measuring AI’s Economic Impacts — Exploratory Grant Proposal","hash":"2a41158412bd21b3014eff7167def6a753ef3a7959e6d68f79466b4c01adba5c"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":0.5418947709343929,"interval":{"lower":-0.14483548454774842,"upper":1.2286250264165341,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":0.5418947709343929,"stddev":0.3503725793276231}},"context":{"rank":21,"percentile":0.4875,"z_score":-0.06965022098214363,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}},{"entity":{"id":"what-happens-if-we-actually-test-it","text":"What Happens If We Actually Test It? — Independent Behavioral Research on Open-Weight AI Models","hash":"b61893259f98682c1e82cf95ec8ca8bd7515bf3358f3ba2ea1680ec61a55f2ae"},"attribute":{"lens":"manifund","axis_key":"epistemic-integrity","axis_prompt":"epistemic integrity of the write-up: honest failure modes, quantified claims, falsifiable milestones","axis_prompt_hash":"1e88f6ebbbd8b052de402122aec2775107a748112fa69f9ee623408c8bf7a77d"},"estimate":{"point":0.8296425005061243,"interval":{"lower":0.19223125229181792,"upper":1.4670537487204307,"coverage":0.95,"method":"normal_approximation"},"distribution":{"family":"normal_approximation","mean":0.8296425005061243,"stddev":0.32520982051750325}},"context":{"rank":9,"percentile":0.7875,"z_score":0.7703691531011544,"cohort_size":40,"scored_at":"2026-08-22T22:19:16.084000Z"},"provenance":{"run_id":"jrun_c6dda3733f59450a892f74f2a30d58a3","model":"openai/gpt-5.6-luna","harness":"cardinal-harness","harness_version":"0.9.0","temperature":0.0,"seed":"18246367922732461848","comparison_budget":320,"comparisons_used":320,"stop_reason":"budget_exhausted","topk_error":5.144060110283961,"run_cost_nanodollars":71830020}}]