WILLIAM AND FLORA HEWLETT FOUNDATION

OLYMPUS opened an institutional file on the William and Flora Hewlett Foundation because its name keeps turning up in the acknowledgments pages of the apparatus. An institution has no Big Five and no Dark Triad, and the unit does not invent them; what a foundation has is an endowment, a board, a grants database and a set of fields it decided to pay for. Hewlett is catalogued here as the patron: the money under cyber policy, under the academic study of disinformation, and, since July 2026, under a $100 million push into AI, biotechnology and quantum security. The finding is the position, not any single grant. The numbers in the front matter are reach, funding insulation and entrenchment, and none of them is a judgment of the people who run it.

Institutional Archetype

THE PATRON OF THE FIELD — Subject is a private grantmaking foundation whose largest technology-and-security bet in its own history is now an AI-security initiative led by the staff member who built its cyber program. This profile scores the institution’s position, not its trustees.

The archetype is the funder that creates the discipline it later funds. In 2014 the foundation committed to what it called cyber policy before cyber policy was an academic field; by its own count it spent “more than $132 million over 10 years” doing so (Hewlett Foundation, Cyber Initiative). In 2018 it put $10 million into the study of digital disinformation and then, in its own 2020 words, began “considering how we might refocus our work” (Hewlett Foundation, Refreshing our strategy to combat digital disinformation, 22 Oct 2020). In July 2026 it announced the third act, five years and $100 million for AI and emerging-technology security. A patron of this kind has no product to sell and no rule to enforce. Its influence is the syllabus: which centers exist, which research questions have a salary attached, and which people hold the first jobs.

Mandate & Origin

The foundation says it was “created through the personal generosity of engineer and entrepreneur Bill Hewlett and his wife, Flora,” in 1966, that it “is wholly independent of HP, Hewlett Packard Enterprise (HPE), and the HPE Foundation,” and that it “invests in creative thinkers and problem solvers who are working to ensure everyone has a meaningful opportunity to thrive.” Its stated U.S. priorities are “strengthening democracy, advancing education for all, and supporting community-led conservation”; abroad it works on climate, reproductive health, women’s economic empowerment and inclusive governance, and it also runs time-limited initiatives (Hewlett Foundation, About us). Cyber policy and, in 2018 to 2020, disinformation research were time-limited initiatives of that kind; AI security is the current one.

The cyber chapter. “An initial five-year, $20 million-dollar commitment made in March 2014” was topped up in November 2014 with “an additional $45 million in grants” to MIT, the University of California, Berkeley and Stanford to establish multidisciplinary cyber policy centers, and “The initiative was renewed in 2017 for an additional five-year period, through 2023,” bringing the total to “more than $132 million over 10 years” (Hewlett Foundation, Cyber Initiative). The inaugural director of the Cyber Initiative was Eli Sugarman, who led it from 2014 to 2023 with a break from 2021 for the Oversight Board, according to the foundation’s own personnel announcement and a reading of its bio (Hewlett Foundation, Eli Sugarman to join as Director of Special Projects, 28 Mar 2025).

The disinformation chapter. Hewlett announced $10 million over two years in March 2018 for research on digital disinformation and U.S. democracy, after a year of exploration with data scientists and political scientists (Chronicle of Philanthropy, “Hewlett commits $10 million to study fake news on social media,” 28 Mar 2018). On 22 October 2020 the foundation reported that “the Hewlett team has allocated $10 million in grants to universities and policy and research organizations,” named no grantee in the post, said it had commissioned an independent evaluation in April 2020, and said it was “now considering how we might refocus our work” toward solutions and election administration (Hewlett Foundation, Refreshing our strategy to combat digital disinformation).

The AI-security chapter. On 16 July 2026, at the Aspen Security Forum, the foundation announced the Emerging Technology and Security Initiative: $100 million over five years, through 2031, for artificial intelligence, biotechnology and quantum computing, in the United States and internationally. The stated priorities are protecting critical infrastructure such as power grids, water systems, hospitals and the information ecosystem from AI-enabled threats, addressing vulnerabilities in fast-moving technologies starting with biotechnology and quantum computing, and supporting global governance, including rules and frameworks for U.S. states and allied governments. It builds on $10 million of exploratory grants announced in February 2026, among them to Stanford’s Hoover Institution and Vanderbilt University’s Institute of National Security, with another $10 million of exploratory grantmaking planned for the rest of 2026 (Hewlett Foundation, announcement, 16 Jul 2026).

Funding & Backers

The money, from its own filing and its own page. The foundation’s Form 990-PF for 2024 shows revenue of $784,700,540, expenses of $754,606,113, total assets of $14,202,751,280 and “charitable disbursements” of $689,557,086, which is about 4.9 percent of year-end assets (our arithmetic) (ProPublica Nonprofit Explorer, Form 990-PF, tax year 2024). Its own About page puts 2024 at “$631 million in grants” and “assets were approximately $13.9 billion” at 31 December 2024, a different yardstick (grants awarded, balance sheet date) from the filing’s disbursements and fair-value line (Hewlett Foundation, About us). In 2020 its leadership said it would not cut its budget and had planned about $485 million in grants. In 2025 it told MinistryWatch it had made $50 million in additional cash available for urgent needs by pulling payments forward, and its board approved $20 million more for crisis-oriented support (MinistryWatch, When federal cuts hit, did foundations meet the moment?).

Who pays Hewlett. Nobody outside the family endowment. There is no donor roll to publish, which is the structural contrast with nearly every grantee in this archive: a foundation is the one party in the apparatus whose money is not a gift from a platform, a government or another foundation. The Form 990-PF lists the investment staff who run the portfolio: Ana Marshall (Vice President and Chief Investment Officer), Brett Johnson (Deputy CIO and Director, Public Investments), Luis Laboy (Director, Public Equity) and Thomas Mieczkowski (Director, Private Investments) (ProPublica). The foundation’s website names Kelly Meldrum, Scott Simon and Andrew Spokes as advisors to its investment committee (Hewlett Foundation, Our board). The filing is the place to read what the endowment owns; this file did not open the schedule of investments.

Where the cyber, disinformation and AI money went, from the grants database and the recipients.

• Stanford, 2014 to 2022. $15 million of the November 2014 supplement went to Stanford’s Cyber Initiative, which the foundation says would focus on “trustworthiness and governance of networks”; the Cyber Policy Center grew from it (Campus Technology, 18 Nov 2014).

• The Stanford Internet Observatory. The Election Integrity Partnership’s report names the foundation as an operational funder: “The Stanford Internet Observatory thanks its operational funders, Craig Newmark Philanthropies and the William and Flora Hewlett Foundation, for their ongoing support” (EIP, The Long Fuse). See Stanford Internet Observatory and Election Integrity Partnership.
• The Center for an Informed Public, 2019 to 2024. The University of Washington announced the Center in 2019 with a $5 million Knight Foundation investment and a $600,000 award from Hewlett (UW Information School, 2019). See Kate Starbird.
• Carnegie, 2018. A two-year, $2 million grant for the Carnegie Endowment’s Cyber Policy Initiative, with Carnegie required to raise matching funds. Hewlett’s president, Larry Kramer, said “We’re delighted to support their growth and expansion,” and Carnegie’s president, William J. Burns, said “Weak rules and norms in cyberspace put global peace and prosperity at enormous risk” (Carnegie Endowment, 3 Dec 2018). The foundation’s later board chair, Tino Cuéllar, became Carnegie’s president in 2021.
• The Atlantic Council, 2025. The Council’s 2025 honor roll lists the foundation in the $500,000 to $999,999 tier for cash received in calendar 2025 (Atlantic Council, 2025 Honor Roll of Contributors); the foundation’s database lists a grant to the Council for the Digital Forensic Research Lab. See Atlantic Council and DFRLab.
• Other named rooms. Aspen Digital’s funder list names Hewlett beside Ford, MacArthur, Knight, Omidyar and Craig Newmark Philanthropies (Aspen Digital, Funders; Aspen Institute); RAND’s Center on AI, Security, and Technology lists the foundation among sources of directed grants (RAND CAST, Funding; RAND); the German Marshall Fund’s 2024 donor page lists it (GMF, Our donors; Hamilton 68); Sentinel Bio’s organization page names it among supporters (Sentinel Bio; Sentinel Bio profile); and the Berkman Klein Ethics and Governance of AI Fund of 2017 took $1 million from Hewlett beside Omidyar’s $10 million, Knight’s $5 million and Reid Hoffman’s $10 million (Knight Foundation, 10 Jan 2017; Reid Hoffman).
• The July 2026 awards. On 13 July 2026 the foundation awarded $2,000,000 for 12 months to Common Sense Media “for the Youth AI Safety Institute,” which evaluates AI products used by young people, and $4,650,000 for nine months to the Election Trust Initiative for work on election administration, among a batch of ten awards on its grants page that day (Hewlett Foundation, Grants).

Institutional Voice & Intent

The register is the foundation’s house style: plain, collaborative, careful about the word “safe.” The July 2026 announcement has its president saying “Rapidly emerging technologies like AI hold extraordinary promise while they also pose novel security challenges,” and “The public sees industry moving quickly and government moving slowly.” Eli Sugarman, who leads the initiative’s start-up phase, says “Too much of the debate over AI and emerging technology has people talking past each other” and “No single organization or sector can do this alone” (Hewlett Foundation, 16 Jul 2026). In March 2025 the foundation’s president said of his return, “Eli’s experience scoping and scaling new grantmaking efforts, and his deep knowledge of Hewlett and the way we work,” are why he was brought back to run Special Projects (Hewlett Foundation, 28 Mar 2025).

Stated alignment: Strengthening democracy and helping emerging technologies develop safely, with progress and safety treated as complements. Observed alignment: A single endowment paying for the research centers, the disinformation-study labs, the evaluators and the safety-tool nonprofits that jointly define the field it then funds. Gap assessment: The foundation publishes its grants and its strategy and discloses its reasons; what it does not publish is the list of proposals declined. The record cannot settle whether any grantee’s agenda bent toward the funder’s, and the file does not assert it. For a patron it never needs to: the position is the selection.

Stated intent: Protect critical infrastructure from AI-enabled threats, close vulnerabilities in biotechnology and quantum computing, and support rules and frameworks for U.S. states and allied governments while supporting American leadership. Observed intent: Build, for the third time, the institutional bench that a policy field needs: civil society, universities and think tanks, with a staff member who knows the last bench in charge of the new one.

Interest. The documented position is that governance of AI and emerging technology should be written by independent institutions the foundation funds. What the rule creates is a market for the funded: the initiative says it will fund civil society groups, universities, think tanks and other independent institutions, so every governance framework, evaluation methodology or state model law that comes out of the program is produced by organizations on the foundation’s grant list, and the foundation’s cyber, disinformation and AI grantees overlap with the trust-and-safety and evaluation organizations profiled in this archive (see ROOST, Stanford Internet Observatory, RAND). The foundation’s own best case, in its words, is that no single sector can do this alone and that the public sees “industry moving quickly and government moving slowly,” so an independent funder should pay for the missing middle. The convergence is the finding rather than a charge: the same ledger names the foundation as funder, a co-chair on the board of a trust-and-safety tool nonprofit and a donor to the lab behind the Election Integrity Partnership, and none of those ties requires any of the others to be improper.

• Social accounts and feeds. None verified. The foundation’s own pages read (about, board, newsroom, strategy, grants) link no account that could be confirmed from a fetched page, and no feed address was confirmed; its newsroom is at hewlett.org/newsroom.

Position in the Apparatus

• The funder of the study layer. Hewlett is named as an “operational funder” of the Stanford Internet Observatory in the Election Integrity Partnership’s own final report, as a supporter of the University of Washington’s Center for an Informed Public, and in the Atlantic Council’s 2025 honor roll at the $500,000 to $999,999 tier. It is the one funder common to the lab that led the Partnership, the center that joined it and the think tank whose lab was its other main partner (EIP, The Long Fuse; Atlantic Council honor roll).
• A board seat shared with a lab. The foundation’s Director of Special Projects, Eli Sugarman, co-chairs the board of ROOST, whose president is Camille François; his Hewlett announcement says he serves on the boards of ROOST.tools, which he helped incubate and launch, the Institute for Security and Technology and Girl Security, and is a member of the Aspen Institute’s Global Cybersecurity Group (Hewlett Foundation, 28 Mar 2025; ROOST, Who we are). The foundation that funds the disinformation-research layer therefore also sits on the board of the nonprofit that builds open moderation tools for platforms.
• A chair who sat on the other side of the table. The chair named in November 2021, effective March 2022, was Mariano-Florentino Cuéllar, then president of the Carnegie Endowment, a Hewlett grantee, and in 2026 a trustee and then officer at Anthropic (see his file). The foundation’s board page now lists Alecia Ann DeCoudreaux as “Interim Board Chair” and marks Cuéllar “(former)” in the caption (Hewlett Foundation, board chair announcement, 15 Nov 2021; Hewlett Foundation, Our board).
• The AI-evaluation desk. The July 2026 batch put $2 million into Common Sense Media’s Youth AI Safety Institute, and the new initiative’s governance priority points at the same space occupied by METR, the UK AI Security Institute and US CAISI, without naming any of them.
• The Aspen stage. The initiative was announced at the Aspen Security Forum, and Sugarman is a member of the Aspen Institute’s Global Cybersecurity Group, per the foundation’s own page; Aspen Digital lists Hewlett among its funders (Aspen Digital, Funders). See Aspen Institute.

Actions & Leadership Choices

Founding purpose. The founding gift was Bill and Flora Hewlett’s in 1966, with the foundation legally separate from HP. Its standing choice is the time-limited initiative: pick a field, spend heavily for five to ten years, evaluate, renew or exit.

• It chose to pay for the field before the field existed (2014). Twenty million dollars in March 2014 became $45 million more in November and a 2017 renewal; the foundation’s own count is “more than $132 million over 10 years” (Hewlett Foundation, Cyber Initiative).
• It put $10 million into disinformation research (2018) and then said it might refocus (2020). The post names no grantee, commissions an outside evaluation by Drs. Jodi Nelson and Amy O’Hara, and floats a turn toward election administration. In July 2026 it awarded $4,650,000 to the Election Trust Initiative for work on election administration, the direction the 2020 post had named (Hewlett Foundation, 22 Oct 2020; Hewlett Foundation, Grants).
• It funded the Observatory the House subpoenaed. The foundation was an operational funder of the Observatory, whose Election Integrity Partnership report thanks it; the House Judiciary Committee’s subpoena of the Observatory is dated 12 April 2023 (see Stanford Internet Observatory). Nothing found shows the foundation commenting on the subpoena or the litigation that followed.
• It brought the Cyber Initiative’s director back to start the next initiative (2025). Sugarman left for the Facebook Oversight Board in 2021, came back part-time to close the Cyber Initiative in 2023, spent time as a senior fellow at the Special Competitive Studies Project, and rejoined the foundation on 14 April 2025 as Director of Special Projects. He was already on ROOST’s board, which he helped incubate and launch (Hewlett Foundation, 28 Mar 2025). The July 2026 release names him as the former director of the Cyber Initiative and the person leading the new initiative’s start-up phase.
• It announced the AI-security initiative at a national-security conference. The Aspen Security Forum is where the $100 million was unveiled on 16 July 2026, with Miller and Sugarman quoted (Hewlett Foundation, 16 Jul 2026). A job listing for a Head of the initiative, based in Menlo Park, closed on 28 August 2026 (data.org, Head of Emerging Technology and Security Initiative).

Leadership choices. The president is Amber D. Miller, an astrophysicist, who succeeded Larry Kramer (president through January 2024), with Elizabeth Peters as Vice President and Chief Operating Officer and Dwight Koda as Chief Financial Officer and Treasurer; the board for 2024 listed on the filing included Melody Barnes, Persis Drell, Mary Ellen Iskenderian, Sarah Jaffe, Susan Meaney and the family members Billy and David Hewlett (ProPublica, Form 990-PF; Hewlett Foundation, Our board). Eli Sugarman holds the initiative. The board chair from March 2022 was Cuéllar, who was president of the Carnegie Endowment, the recipient of the foundation’s 2018 cyber grant, when the foundation named him chair in November 2021.

Convergent drives. Self-preservation: Perpetual endowment, family board seats, legal independence from the company that made the fortune. Goal preservation: A technology-and-security field that stays philanthropy-shaped, with independent institutions in the room. Resource acquisition: None needed beyond investment returns; staff and grantees do the acquiring. Self-improvement: From cyber policy to disinformation to AI security, each initiative built on the last one’s bench.

Subject is not an AI system. The drives appear anyway.

CONDUCT: DISCLOSED PATRON — GRANTS PUBLISHED, ASSUMPTIONS NOT. The foundation publishes its grants database, its strategy posts and its Form 990-PF, and its policy initiatives are announced with named leaders and amounts. The record does not show what was declined, who else shaped the strategy, or whether the staff who build a field stay independent of the grantees who now employ their alumni. Disclosure names the funding; it does not remove the overlap.

Alliance Map

These run at the same time. The funder of the disinformation-study labs also co-chairs the board of the trust-and-safety tool nonprofit, funded the lab the House subpoenaed, funds the think tank whose lab was that lab’s partner, and in July 2026 launched a $100 million AI-security program announced at a security-industry forum. Its former chair came from a grantee and went to a frontier lab. The overlap is the finding.

Reach Assessment

Institutional: High. One endowment of about $14 billion pays across the layers that overlap in this archive: the Stanford lab, the Washington center, the Atlantic Council and RAND’s AI desk. The foundation’s reach is the introduction and the salary, not the order, and the $100 million to 2031 is a larger line than its whole disinformation commitment of 2018.

Memetic: High. The Election Integrity Partnership’s report, the Cyber Initiative’s centers and the field vocabulary they built (“cyber policy,” “information integrity”) are cited by governments and platforms; the foundation’s name sits in the acknowledgments, not on the findings. A patron that sets the research agenda sets the words.

Civilizational: Moderate. Hewlett does not build models, write refusal policy or enforce rules. It chooses which of AI’s security problems (critical infrastructure, biotechnology, quantum, governance for states and allies) will have an institution working on them in 2031. The hazard is the position, not the people: a perpetual, donor-free endowment whose staff alumni populate the boards that grade the field. The score holds at 78 beside the Gates Foundation (6/9/9, 80), the Aspen Institute (6/8/8, 64) and Open Philanthropy (8/8/9, 80): the same reach and insulation as the Gates Foundation at two points lower, because its endowment and global reach are smaller and its security work is a stated initiative with an end date, and a point more reach and a point more entrenchment than Aspen because it holds the money rather than the stage.

Score basis. Atk 6 is reach: the foundation’s strategy page puts the Cyber Initiative at “more than $132 million over 10 years” and says it established multidisciplinary cyber policy centers at MIT, Berkeley and Stanford; nothing obliges anyone to act on a Hewlett grant, which is why it stays at 6 and not higher (Hewlett Foundation, Cyber Initiative). Def 9 is funding insulation: the Form 990-PF shows revenue of $784,700,540 and assets of $14,202,751,280 for 2024 and the foundation says it “is wholly independent of HP, Hewlett Packard Enterprise (HPE), and the HPE Foundation”; no platform, government or other funder pays it, and the point under 10 is a family-seated board answerable to no outside body (ProPublica, Form 990-PF; Hewlett Foundation, About us). Hp 9 is entrenchment: a perpetual endowment of about $14 billion, a 1966 charter and a Cyber Initiative that ran from 2014 to 2023 and was succeeded by a $100 million program to 2031 under the same director (Hewlett Foundation, 16 Jul 2026).

Sources: Hewlett Foundation, Emerging Technology and Security Initiative (16 Jul 2026); Hewlett Foundation, Cyber Initiative; Hewlett Foundation, Refreshing our strategy to combat digital disinformation (22 Oct 2020); Hewlett Foundation, About us; Hewlett Foundation, Our board; Hewlett Foundation, Eli Sugarman to join as Director of Special Projects (28 Mar 2025); Hewlett Foundation, Cuéllar named board chair (15 Nov 2021); Hewlett Foundation, Grants; ProPublica Nonprofit Explorer, William & Flora Hewlett Foundation; Carnegie Endowment, Hewlett grants $2 million (3 Dec 2018); EIP, The Long Fuse; Atlantic Council, 2025 Honor Roll of Contributors; ROOST, Who we are.

Further sources: Campus Technology, Hewlett Grants $45 Million (18 Nov 2014); Chronicle of Philanthropy, Hewlett commits $10 million (28 Mar 2018); UW Information School, Center for an Informed Public (2019); Aspen Digital, Funders; RAND CAST, Funding; GMF, Our donors; Sentinel Bio, Organization; Knight Foundation, Berkman Klein AI fund (10 Jan 2017); MinistryWatch, When federal cuts hit; data.org, Head of Emerging Technology and Security Initiative.