Truace
Subscribe
The Index recomputed live from the record

What the evidence says.What the public feels.

The record holds 939 sourced gains and 772 sourced problems, averaging 68 and 66 on the index score. Readers have logged 22 public signals on the Pulse, which is kept apart and never counted as evidence.

939 gains
772 problems
Every sourced claim in the Index, one square each, shaded by the strength of its evidence. High Moderate Emerging

Ranks distinct AI gain and problem claims from the published record. Scores reward impact, independent source strength, scale, confidence, and recency.

1,711 results
Show filters and sorting

Download every matching row, not just this page:Export CSVExport JSON

AI gains · 939

521
CrimeStableModerate evidence · 1 source

Frontier chatbots showed less concerning behavior in newer models and when early escalation interventions were applied during mental-health conversations.

On 2026-08-07, Nature Medicine published a clinically validated auditing framework called SIM-VAIL that simulates users with psychiatric vulnerabilities to test frontier chatbots including Claude, ChatGPT, Gemini, Grok and Llama. Across 810 multi-turn conversations with 30 simulated profiles and scoring on 13 risk dimensions, the study observed widespread concerning behavior that accumulated over turns.

Impact 30%
49
Evidence 25%
95
Scale 20%
35
Confidence 15%
87
Recency 10%
88

Updated Aug 10, 2026 · TRV-2026-0726

67Index score
522
HealthStableModerate evidence · 1 source

AI augmentation of operational workflows under human oversight improves oncology trial feasibility and patient identification, with tools for enrollment screening and monitoring now implemented at select cancer centres.

On 2026-08-07, a Review in Nature Reviews Clinical Oncology described how AI enabled by electronic health record datasets and machine learning is being applied across pre-trial design, conduct, and post-trial inference in oncology. It reported that the most immediate evidence-supported uses are operational workflows under human oversight, including patient identification, eligibility assessment, data extraction, and trial monitoring, now implemented at select cancer centres.

Impact 30%
49
Evidence 25%
95
Scale 20%
35
Confidence 15%
87
Recency 10%
88

Updated Aug 10, 2026 · TRV-2026-0725

67Index score
523
HealthStableModerate evidence · 1 source

Researchers applied a large language model to analyze 125 compensated orthodontic cases and identified that most harms were multifactorial and largely avoidable, creating a foundation for future safety initiatives.

Researchers retrospectively reviewed 125 orthodontic claims approved for compensation by the Danish Dental Compensation Association from September 2019 to August 2024. They applied Eindhoven incident analysis and AI-assisted qualitative analysis using a large language model to identify root causes and rate perceived avoidability on a 6-point scale.

Impact 30%
49
Evidence 25%
95
Scale 20%
35
Confidence 15%
87
Recency 10%
87

Updated Aug 9, 2026 · TRV-2026-0717

67Index score
524
ScienceStableModerate evidence · 1 source

An ensemble combining Random Forest and Multi-Layer Perceptron improved genomic prediction of residual feed intake in 220 UK Holstein cows to R2=0.39 and RMSE=0.086, outperforming conventional gBLUP.

Using genomic data from 220 UK Holstein cows, researchers tested Random Forest and Multi-Layer Perceptron models against conventional gBLUP for predicting residual feed intake, a feed-efficiency trait. The ensemble of RF and MLP achieved the best reported performance with R2=0.39 and RMSE=0.086, while SHAP analysis identified distinct and overlapping candidate genes.

Impact 30%
49
Evidence 25%
95
Scale 20%
35
Confidence 15%
87
Recency 10%
87

Updated Aug 9, 2026 · TRV-2026-0714

67Index score

AI problems · 772

521
HealthStableModerate evidence · 1 source

National EHR networks currently function primarily as research platforms, with few ML/AI models prospectively evaluated or integrated into workflows due to heterogeneous capture, privacy constraints, limited representativeness, and need for local recalibration.

Researchers conducted an environmental scan through September 2025 of 23 US national EHR networks that aggregate patient-level data, ranging from under 1 million to over 200 million patients, and reviewed 34 ML/AI studies built on them. Most networks used common data models, yet few models were prospectively evaluated or integrated into clinical workflows.

Impact 30%
49
Evidence 25%
95
Scale 20%
35
Confidence 15%
87
Recency 10%
82

Updated Jul 13, 2026 · TRV-2026-0187

67Index score
522
HealthStableModerate evidence · 1 source

Students in Qatar reported low-to-moderate trust in AI-based mental health tools and concerns about loss of human interaction, overreliance on technology, and diagnostic accuracy.

On 2026-05-06 a peer-reviewed survey study reported results from 220 university students in Qatar about AI in mental health support. Students reported low-to-moderate awareness and trust, said they were prepared to use AI for stress management but did not want it to replace face-to-face therapy, and cited benefits of cost reduction and 24/7 accessibility alongside concerns about human interaction, overreliance and diagnostic accuracy.

Impact 30%
49
Evidence 25%
95
Scale 20%
35
Confidence 15%
87
Recency 10%
82

Updated Jul 13, 2026 · TRV-2026-0184

67Index score
523
HealthStableModerate evidence · 1 source

Fragmented AI governance frameworks that assume extensive resources create barriers to adoption for smaller healthcare organizations.

On 2026-02-11, authors reported a systematic review of 35 healthcare AI implementation frameworks from 2019-2024, identifying seven critical governance domains. They used those findings to develop HAIRA, a five-level maturity model from Level 1 Initial/Ad Hoc to Level 5 Leading with benchmarks across the domains.

Impact 30%
49
Evidence 25%
95
Scale 20%
35
Confidence 15%
87
Recency 10%
82

Updated Jul 13, 2026 · TRV-2026-0182

67Index score
524
PolicyStableModerate evidence · 1 source

Current defensive systems remain siloed and lack integrated semantic reasoning across domains and languages, failing to correlate technical cyber indicators with coordinated narrative manipulation.

Published May 9, 2026, this scoping review assessed OSINT, SOCMINT, and NLP tools for hybrid-threat detection against operational requirements drawn from Russian and Chinese military tradecraft and European operational experience. It found individual disciplines technically advanced but defensive systems siloed, identifying a persistent semantic gap in cross-domain and cross-language reasoning.

Impact 30%
49
Evidence 25%
95
Scale 20%
35
Confidence 15%
87
Recency 10%
82

Updated Jul 13, 2026 · TRV-2026-0178

67Index score

Recomputed live from the record · Oct 13, 2026, 4:42 PM