evaluator; distance from a lab: 3
Center for AI Safety
Scorecard
Hazard benchmarks (WMDP, HLE); director advises xAI.
Nonprofit, San Francisco, US. Confidence med. Domains: benchmarks, misuse. Scores by preset: Lab procurement 60; Regulator or auditor 60; Public trust 60; Equal weights 60. Weakest dimension: personnel 1/4.
What would move the score. A public recusal policy covering the xAI relationship.
Funding 3/4
Mostly philanthropic or public money; some lab-linked pooled funds.
evidence: tier 4 (press)
Anchor 3 on funding given the cited signals.
- Funded mainly by SFF rather than lab-linked pools.
It looks like there are some good funding opportunities in AI safety right now (80,000 Hours, 2025-10-13) tier 4 pressconfirmed2026-09-14
Governance 3/4
Nonprofit or public body with a COI policy.
evidence: tier 3 (self)
Anchor 3 on governance given the cited signals.
- Nonprofit.
Center for AI Safety (CAIS, 2026) tier 3 selfconfirmed2026-09-14
Personnel 1/4
Leaders hold equity or advisory roles at labs; informal recusal.
evidence: tier 3 (self)
Anchor 1 on personnel given the cited signals.
- Director is a safety adviser to xAI while the organization's benchmarks are used to grade xAI models.
It looks like there are some good funding opportunities in AI safety right now (80,000 Hours, 2025-10-13) tier 4 pressconfirmed Center for AI Safety (CAIS, 2026) tier 3 selfconfirmed2026-09-14
Access depth (lab-granted) 2/4
Pre-release with safeguards off or extended time.
evidence: tier 3 (self)
Anchor 2 on access given the cited signals.
- Public-model access.
Center for AI Safety (CAIS, 2026) tier 3 selfconfirmed2026-09-14
Scope control 3/4
Evaluator sets scope and can add questions.
evidence: tier 3 (self)
Anchor 3 on scope given the cited signals.
- Sets own benchmark design.
Center for AI Safety (CAIS, 2026) tier 3 selfconfirmed2026-09-14
Publication rights 3/4
Publishes; redaction limited to security; redactions disclosed.
evidence: tier 3 (self)
Anchor 3 on publication given the cited signals.
- Publishes results.
Center for AI Safety (CAIS, 2026) tier 3 selfconfirmed2026-09-14
Method transparency 4/4
Open code, tasks, reproducible runs, factsheets.
evidence: tier 3 (self)
Anchor 4 on methods given the cited signals.
- Open benchmarks widely used.
Center for AI Safety (CAIS, 2026) tier 3 selfconfirmed2026-09-14
Role incompatibility 3/4
Tools are open or free to the ecosystem.
evidence: tier 3 (self)
Anchor 3 on products given the cited signals.
- Co-produced Humanity's Last Exam with Scale AI, a Meta-owned vendor.
Center for AI Safety (CAIS, 2026) tier 3 selfconfirmed2026-09-14
Traced money and ties
hop 2: 2 rows (recommendation $1.4M). Confirmed rows: 2 of 2. Dollar sums: confirmed + unaudited USD rows only; imported figures are not re-derived and never summed. Second hop traced for 1 of 1 sources. Ties: none within two steps recorded. Bounded negatives on file: 0.
Ledger
Money in
- T57 from Survival and Flourishing Fund: recommendation $1.10M, 2023-2025. SFF to CAIS, about $1.1M confirmedtier 4 presshttps://80000hours.org/2025/01/it-looks-like-there-are-some-good-funding-opportunities-in-ai-safety-right-now/
- T71 from Survival and Flourishing Fund: recommendation $0.29M, 2025. SFF 2025: $289,000 to CAIS; the CAIS Action Fund is a separate entity ($772,000, not added) confirmedtier 3 selfhttps://survivalandflourishing.fund/2025/recommendations
Roles hosted
- R43 Dan Hendrycks: principal. director confirmedhttps://safe.ai/
One hop away
Funding graph, focused
Neighbours at full strength, everything else faded. Hover any name to move the focus; click a name to open its page.