Independent Assessment
AI Safety
Scorecard
40 indicators. 6 frameworks. Zero self-reporting.
Scoring startups, deployers, and frontier labs on safety practices using independently verifiable evidence.
Companies
16
scored
Avg Score
50.4
out of 100
Indicators
40
per company
Frameworks
6
cross-mapped
Safety Leaderboard
Ranked by overall safety score. Click any row for full breakdown.
#
Company
Grade
GOV
TECH
RISK
REG
ENG
EU AI Act
1
AnthropicUS
foundation modelsai safety
A-
72.878
74
68
65
82
GPAI Systemic
2
AI Underwriting CompanyUS
ai governanceai safety tooling
A-
70.072
68
70
75
60
Limited
3
Gray SwanUS
ai securityred teaming
B+
64.755
78
72
45
80
Limited
4
OpenAIUS
foundation modelsenterprise ai
B
59.162
65
52
55
60
GPAI Systemic
5
Haize LabsUS
red teamingai safety tooling
B
56.748
70
62
40
68
Limited
6
GoodfireUS
interpretabilityai safety
B
56.355
72
55
30
75
Limited
7
MusubiUS
content moderationtrust and safety
B-
48.552
48
45
55
35
Limited
8
Patronus AIUS
ai safety toolingevaluation
C+
47.642
60
45
38
55
Limited
9
Confident SecurityUS
ai privacyconfidential computing
C+
47.045
55
35
50
50
Limited
10
Lucid ComputingUS
ai governanceexport controls
C+
46.040
50
45
55
35
Limited
11
Andon LabsUS
autonomous agentsai safety
C+
44.545
55
50
20
55
Limited
12
MetaUS
foundation modelsopen source ai
C
43.840
52
35
45
48
GPAI
13
Red Queen BioUS
biosecuritybiodefense
C
43.048
42
55
25
45
Limited
14
TheoremUS
formal verificationcode safety
C
41.335
62
40
25
40
Limited
15
SoftmaxUS
alignment researchai safety
C
41.350
45
40
20
55
Limited
16
ArchiaUS
ai infrastructureagent deployment
D-
23.325
30
15
25
15
Limited
Need deeper analysis?
Get indicator-level evidence, trend tracking, and custom peer benchmarking for your AI portfolio due diligence.
Scores reflect publicly available information only. A low score may indicate limited transparency rather than poor safety practices.
Mappera is not affiliated with any AI company assessed. Methodology v0.3 — updated quarterly.