These visualizations are experimental and still being refined. Designs and data presentation may change.

Dataset Coverage by Subdomain

How records distribute across the 24 risk subdomains for each domain-taxonomy dataset. Color intensity is normalized per column, so darker cells indicate more records relative to that dataset’s maximum. Mitigations use their own taxonomy and are not included.

Click column headers to sort. Click any cell to browse its filtered records.

Subdomain	Risks	Incidents	Governance Score
1.1 Discrimination	83 (5.5%)	118 (8.1%)	192.5 (5.2%)
1.2 Toxic content	116 (7.7%)	90 (6.2%)	77.0 (2.1%)
1.3 Unequal performance	17 (1.1%)	34 (2.3%)	150.5 (4.1%)
2.1 Loss of privacy	80 (5.3%)	88 (6.0%)	277.5 (7.5%)
2.2 AI security vulnerabilities	112 (7.4%)	21 (1.4%)	345.5 (9.3%)
3.1 False information	53 (3.5%)	187 (12.8%)	165.5 (4.5%)
3.2 Loss of consensus reality	22 (1.5%)	4 (0.3%)	63.0 (1.7%)
4.1 Disinformation & influence	84 (5.6%)	135 (9.3%)	193.0 (5.2%)
4.2 AI weapons & cyberattacks	82 (5.4%)	13 (0.9%)	177.0 (4.8%)
4.3 AI fraud & scams	77 (5.1%)	394 (27.0%)	203.0 (5.5%)
5.1 Overreliance & unsafe use	60 (4.0%)	36 (2.5%)	150.0 (4.0%)
5.2 Loss of human agency	46 (3.0%)	3 (0.2%)	125.5 (3.4%)
6.1 Power centralization	54 (3.6%)	6 (0.4%)	28.5 (0.8%)
6.2 Inequality & unemployment	55 (3.6%)	6 (0.4%)	55.5 (1.5%)
6.3 Devaluation of human creativity	31 (2.1%)	5 (0.3%)	22.0 (0.6%)
6.4 Competitive dynamics	20 (1.3%)	3 (0.2%)	203.0 (5.5%)
6.5 Governance failure	61 (4.0%)	3 (0.2%)	365.0 (9.9%)
6.6 Environmental harm	57 (3.8%)	1 (0.1%)	42.0 (1.1%)
7.1 AI misalignment	100 (6.6%)	3 (0.2%)	76.5 (2.1%)
7.2 Dangerous capabilities	77 (5.1%)	0 (0.0%)	140.0 (3.8%)
7.3 Capability & robustness	126 (8.3%)	302 (20.7%)	330.0 (8.9%)
7.4 Transparency & interpretability	42 (2.8%)	5 (0.3%)	308.0 (8.3%)
7.5 AI welfare	3 (0.2%)	0 (0.0%)	2.5 (0.1%)
7.6 Multi-agent risks	53 (3.5%)	0 (0.0%)	11.0 (0.3%)

Key Takeaways

1.7.3 Capability & robustness has the most combined records (899 total) across risks, incidents, governance documents.
2.The top 5 subdomains account for 36% of all risk entries, indicating moderate concentration.

Mitigations use their own classification taxonomy and are not shown here. Expert Concern shows the percentage of survey respondents who ranked a subdomain in their top-3 concerns.