2025-11-17T15:11:35-08:00 2024-12-31 872635810 Tovella Dowling PC 501 W Broadway Ste 1540 San Diego CA 92101 990 2024-01-01 881751310 Center for AI Safety Inc CENT 5102105245 555 Montgomery Street Suite 1501 San Francisco CA 94111 Oliver Zhang Director 2025-11-17 true true Julie U Nguyen Esq P01881055 6199309332 2024 2025-03-06 01:10:19Z Daniel Hendrycks 10769815 false X www.safe.ai X 2022 CA to reduce societal-scale risks associated with artificial intelligence AI by advancing safety research, building the field of AI safety researchers, and promoting safety standards. 3 2 25 2 0 15594432 9596669 256598 484973 333081 9786 48737 16089191 10235085 0 0 1857974 2830523 0 145689 6276475 4333084 8134449 7163607 7954742 3071478 13989499 12638754 5451259 1029036 8538240 11609718 X to reduce societal-scale risks associated with artificial intelligence AI by advancing safety research, building the field of AI safety researchers, and promoting safety standards. false false 2947497 256598 Compute Cluster: CAIS runs an initiative to provide free compute for research projects in ML safety, based on a cluster of 256 A100 GPUs with dedicated support staff to enable non-industry researchers to conduct advanced AI safety research requiring cutting-edge models. Since launching in 2022, the CAIS Compute Cluster has become a cornerstone of global AI safety research. By 2024, more than 400 researchers had used the cluster to conduct safety-focused ML research, generating 121 research papers and accumulating over 5,000 citations. The cluster continues to remove financial and technical barriers for non-industry researchers, enabling cutting-edge experimentation through free access to compute. 1168082 Research: CAIS conducts foundational research aimed at mitigating societal-scale risks posed by artificial intelligence. We develop technical benchmarks and methods which improve AI safety. We also examine AI safety from a multidisciplinary perspective that incorporates safety engineering, complex systems, international relations, and other disciplines. In 2024, CAIS made major strides in foundational AI safety by developing new tools to measure and mitigate national-security risks posed by advanced models. This work included releasing the WMDP Benchmark, the first public dataset for evaluating hazardous knowledge in AI systems; creating circuit breakers, a defense mechanism that reliably prevents harmful outputs even under adversarial attacks; and coordinating Humanity's Last Exam, a global collaboration of nearly 1,000 contributors from more than 500 institutions across 50 countries to assess frontier-level reasoning across 3,000 expert-level questions. CAIS researchers also published 8 papers over the year. 874948 777874 Advocacy: CAIS works to bring AI safety research into practice through engagement with industry leaders, policymakers, and research labs, aiming to build awareness and establish standards for safe AI development and deployment. CAIS saw significant growth in public engagement, expanding its AI Safety Newsletter to 24,000 subscribers, a nearly threefold increase from the prior year, and contributing to over 100 media stories in outlets such as The New York Times, The Wall Street Journal, and The Washington Post. Through research translation, public communication, and collaboration with academic and industry partners, CAIS helped elevate AI safety as a mainstream concern and provided technical grounding for ongoing policy discussions at both state and federal levels. 520385 Field Building: CAIS builds and strengthens the AI safety research community through various initiatives designed to provide support and lower barriers to entry. In 2024, CAIS expanded its field-building initiatives by making AI safety education more accessible and interdisciplinary. We published the textbook Introduction to AI Safety, Ethics & Society, and launched an accompanying online course that enrolled 240 students from varied academic backgrounds. CAIS also strengthened community engagement by hosting workshops at major conferences including NeurIPS, where our workshop on agentic AI systems accepted 34 papers, and organizing AI safety socials at ICML and ICLR with 200+ researchers participating. The SafeBench competition, supported by Schmidt Sciences, attracted nearly 90 benchmark submissions to advance empirical AI safety evaluations. 1298259 6288786 true true false true false false false false false false true false false false true true false true false false false false false false false false false false false true false false false false false false false false false false false false false false false false false true 27 0 true 25 true false false false false false false false 0 false false false false false false false X 3 2 false false false false false false false true true false false true true true true true true true true false CA FL MA NY X Accomptant Inc 5102105245 12630 FREEMAN AVENUE HAWTHORNE CA 90250 Daniel James Hendrycks Executive Dir. 40.00 0.00 X X 314534 0 6749 Nicholas Beckstead Policy Lead 40.00 0.00 198071 0 14458 Steven Basart Research engineer 40.00 0.00 171272 0 20849 Bri Treece COO 40.00 0.00 X 169980 0 12095 Oliver Zhang Managing Director Director 40.00 0.00 X 155573 0 12251 William Hodgkins Project Manager 40.00 0.00 150379 0 13134 Long Phan Research Engineer 40.00 0.00 143070 0 9584 Alice Gatti Research Engineer 40.00 0.00 131032 0 12691 Rune Tybirk Kvist Director 1.00 0.00 X X 0 0 0 Jaan Tallin Director 1.00 0.00 X 0 0 0 1433911 101811 8 false true false 0 9596669 9596669 Computing Services 900099 256598 256598 256598 384256 384256 0 0 0 483555 534730 -51175 -51175 -51175 0 0 0 Reimbursements 900099 48737 48737 48737 10235085 256598 381818 0 0 0 0 671182 460536 124602 86044 0 1825944 1597557 206150 22237 59218 52914 5625 679 94909 80146 12223 2540 179270 145616 25498 8156 0 106916 106916 39769 39769 0 0 0 548079 455553 92526 15291 15291 9814 7972 1396 446 0 0 212859 172899 30276 9684 55157 44803 7845 2509 0 0 0 0 2615 2124 372 119 8581 8581 compute cluster 2947725 2947497 228 in office meals 110221 89530 15677 5014 software 91744 74521 13049 4174 website 61249 49751 8712 2786 123064 107367 14396 1301 7163607 6288786 729132 145689 1286542 10026659 0 190000 1875682 25262 125828 0 0 0 0 27965 98174 17098 4792 10352 12306 12449378 0 0 0 0 500105 13989499 12638754 191457 545211 5259802 483825 5451259 1029036 X 7768032 10933932 770208 675786 8538240 11609718 13989499 12638754 3071478 X false true X true false X 1398698 15594432 9596669 26589799 0 0 1398698 15594432 9596669 26589799 13253473 13336326 47 484973 333081 818101 0 1513 9786 48737 60036 27467936 256598 X 0.00000 Support Schedule: Additional Supplemental Information Part II, Sec. A, Line 1, column d 2022 Gifts, grants, contributions, and membership fees received was adjusted to reflect a partial refund of a 2022 grant in the amount of 5,259,802. RESTRICTED RESTRICTED RESTRICTED RESTRICTED RESTRICTED RESTRICTED RESTRICTED RESTRICTED false false 56346 40898 97244 7066363 7163607 508180 127045 false 556722 508180 1064902 1597353 97244 97244 139181 127045 266226 399339 56346 56346 4568 108 4460 12530 4684 7846 12306 Operating Lease Liability 483825 483825 X 10235085 10235085 10235085 7163607 7163607 7163607 Part X : FIN48 Footnote The Corporation is a public charity, and is exempt from income taxes under Section 501(c)(3) of the Internal Revenue Code and Section 23701(d) of the California Revenue and Taxation Code. The Corporation believes that it has appropriate support for any tax positions taken, and, as such, does not have any uncertain tax positions that are material to the financial statements. The Corporation is not a private foundation. The Corporation?s Return of Organization Exempt from Income Tax for the years ended December 31, 2024, 2023 and 2022 are subject to examination by the Internal Revenue Service and State taxing authorities, generally three-to-four years after the returns were filed. X X X X X false false false false false false false false true true Bri Treece COO 169980 6815 5280 182075 Daniel James Hendrycks Executive Dir. 314534 6749 321283 Long Phan Research Engineer 143070 3200 6384 152654 Nicholas Beckstead Policy Lead 198071 7808 6650 212529 Oliver Zhang Managing Director Director 155573 6223 6028 167824 Steven Basart Research engineer 171272 7200 13649 192121 William Hodgkins Project Manager 150379 6038 7096 163513 Part I, Line 8: Amounts reported on 990 VII pursuant to initial contract exemption described in Regs The organization hired the Chief Operating Officer and the highest compensated employees (together listed as employees #3 - #7) pursuant to written employment contracts. At the time of hire, these individuals were neither directors, officers, nor otherwise disqualified persons with respect to the organization and were not in positions to exercise substantial influence over its affairs. The contracts' terms were fixed and specified all elements of compensation including salary, benefits, and any potential bonus arrangements. The contracts were negotiated at arm's length before the individuals began providing services to the organization. The organization's board reviewed comparable compensation data for similar positions at comparable organizations before approving the contract terms. Since these agreements represent the initial contracts between the organization and these employees, and the employees were not previously in positions to exercise substantial influence over the organization's affairs, the fixed payments under these contracts are subject to the initial contract exception under Treasury Regulations section 53.4958-4(a)(3). Form 990, Part VI, Section B, Line 11b A copy of the Form 990 was provided to the officers and directors for review and comment prior to filing. Form 990, Part VI, Section B, Line 12c Upon engagement, election, or hiring, directors, officers, and key employees agree to comply with the conflict of interest policy and to disclose any conflicts of interest, which are then handled in accordance with the policy. Form 990, Part VI, Section B, Line 15a It is the organization's policy that compensation of any director, officer, or key employee is reasonable. Determination of reasonableness entails gathering relevant market data followed by review, deliberation, and written approval by a disinterested board. Form 990, Part VI, Section B, Line 15b See response to Form 990, Part VI, Line 15a. Form 990, Part VI, Section C, Line 18 The organization's Form 1023 and Form 990 returns are accessible to the public through the California Attorney General Charities Registry. The organization can also make such documents available upon written request. Form 990, Part VI, Section C, Line 19 Documents subject to public disclosure or inspection are accessible through the California Attorney General Charities Registry or can be made available upon written request to the organization.