CalData · California Office of Data and Innovation

Public data, public trust

Data ethics and governance in California state government

Lucy Andrews

October 12, 2026

Today’s agenda

  • About me 5 minutes
  • Lecture: public data, public trust 35 minutes
  • Exercise: to publish or not to publish 30 minutes
  • Debrief 10 minutes

About me

Hi, I’m Lucy

Macalester College shield

BA in Geology, Macalester College

Water drop icon

Drinking water efficiency work

UC Berkeley logo

PhD in Environmental Science, Policy, and Management

California Department of Water Resources seal

Senior environmental scientist

Office of Data and Innovation logo

Senior data governance strategist

Deliver better government services to the people of California through technology and design.

ODI was created to deliver better services to Californians

Collaborate with partners

Work with agencies, departments and the Governor’s Office with the support of the Legislature.

People not tech first

Focus first on people’s needs then find the right technology and policy solutions to address those needs.

Serve Californians

Rethink, rebuild, and create services and programs that better serve and engage Californians.

Ship solutions

Deliver insights, strategy, websites, digital services, data solutions, training and guidance.

A request lands on your desk

The dilemma

A state department has county-level enrollment counts for a benefits program, broken out by race/ethnicity and age group.

A journalist has filed a Public Records Act request. Additionally, an advocacy group wants the table posted on the State’s open data portal, data.ca.gov.

Some cells count only a handful of people. Many of those are American Indian and Alaska Native residents of small rural counties (using census terminology).

What would you do?

A. Publish as is. The Public Records Act presumes access, and these are counts, not individual records.

B. Suppress small cells. Hide counts under 11, plus any cells that would let someone recover them, then publish.

C. Aggregate. Combine small counties into regions, or small groups into broader categories, so no cell is small.

D. Consult first. Hold the release until affected tribes have been consulted, then decide.

E. Withhold. Deny the request and keep the table off data.ca.gov, arguing that the privacy risk clearly outweighs the public interest in disclosure.

Go to menti.com and enter code 9406 1118
Only the first 50 answers will be logged, so chat with a neighbor and submit one answer for your pair!

QR code linking to the Mentimeter poll

Public data vs. private data

Why the ethical bar is different for government than industry

Why government data is different

No exit

You can delete an app. You can’t opt out of Medi-Cal eligibility checks, the DMV, or taxes. Collection happens under legal compulsion or real need.

Public accountability

Agencies answer to voters, the Legislature, courts, and the State Auditor, not to shareholders.

Duty to disclose

Government records are presumed open. Companies can protect data as trade secrets.

Buy, not build

Much of the state’s technology is procured. Contracts are where values get enforced.

Same state, different rules

Consideration Businesses State agencies
Privacy law CCPA, as amended by the CPRA (Prop 24, 2020) Information Practices Act of 1977
Core rule Consumer rights to know, delete, correct, and opt out of sale Keep only information “relevant and necessary” to an authorized purpose
Disclosure Trade secrets protected California Public Records Act: access is a “fundamental and necessary right”
Enforcement California Privacy Protection Agency and Attorney General Agencies, courts, State Auditor, Legislature

Watch out: two CPRAs

CPRA means both the California Privacy Rights Act (businesses) and the California Public Records Act (government).1

“In framing a government which is to be administered by [people] over [people], the great difficulty lies in this: you must first enable the government to control the governed; and in the next place oblige it to control itself.”

James Madison, Federalist No. 51 (1788)

“The right of privacy is the right to be left alone. It is fundamental and compelling.”

George Moscone and Kenneth Cory, ballot argument for Proposition 11, which added privacy to California’s Constitution (1972)

Data is where this social contract bargain plays out today: the state collects what it needs to govern and serve, and owes the people it governs limits and accountability in return.

What the social contract lens adds

  • Government data collection is part of a bargain: people grant the state the power to collect, and the state owes them limits and accountability.
  • The same data that delivers a benefit can also be used to deny, police, or surveil.
  • So the questions aren’t just is it accurate? and is it secure? but for whom, by whom, and for what?

Key takeaway

In government, collecting data is a use of state power, so the ethical bar is often higher than in the private sector.

Transparency vs. privacy

Two public goods in tension

Open by default…

  • Open data supports accountability, research, journalism, and better services.
  • data.ca.gov hosts thousands of state datasets from dozens of organizations.1
  • The California Open Data Policy (2019) sets the default.2
  • But open data excludes personal and confidential information, and each dataset should be reviewed against de-identification guidelines first.3

…but minimize by design

  • Collect less. The IPA limits agencies to information “relevant and necessary” to an authorized purpose.1
  • Disclose carefully. Agencies may not disclose personal information in a way that links it back to the individual, with listed exceptions.
  • Beware the mosaic effect. Releases that look harmless on their own can identify people once combined.

How easy is re-identification?

In 1990 Census data, an estimated 87% of the U.S. population was likely unique on just ZIP code, gender, and date of birth.2

Data minimization in practice: ODI’s toolkit

Data minimization

“Collecting and keeping the information you need and nothing more.”1

Ask before you collect

“Do I really need this information to fulfill my goal?” and “Will I actually use it to deliver a service or meet a legal requirement?” If not, don’t collect or store it.

Purpose limitation

Use data only for the reason it was collected, unless there’s proper justification or consent.

Minimum necessary

Access or share only what the task needs. For example, an eligibility worker doesn’t need a full medical history.

ODI’s data minimization toolkit grounds these principles in the Information Practices Act, and the Statewide Data Strategy pairs open data with privacy by design and data minimization.2

The small numbers problem

California’s health and human services agencies use shared de-identification guidelines.1

  • Counts based on fewer than 11 people can trigger suppression.
  • Complementary suppression: hide extra cells so the hidden ones can’t be recovered from totals.
  • Rates and percentages built on suppressed cells get suppressed too.

The tension

Suppression protects individuals, but it can erase whole communities from the public record—often the smallest, most rural, and/or most marginalized.

“What gets counted counts.”

Joni Seager, as quoted in Catherine D’Ignazio and Lauren Klein, Data Feminism (2020)

Case: CalGang

CalGang is a shared gang database used by law enforcement across the state. In 2016, the State Auditor reviewed it.1

42

people were listed as younger than 1 year old when entered

13 of 100

individuals reviewed had no adequate support for being in the database

250+

people had purge dates more than 100 years in the future

What changed: AB 90 (2017) put CalGang in statute under Department of Justice oversight.2 Poor data governance had become a civil rights problem.

ODI in practice: Engaged California

A state deliberative-democracy platform, launched in 2025 with Los Angeles wildfire recovery, and statewide in 2026 on AI and work.1

  • Participation is anonymous, and names aren’t revealed.
  • Comment data is published openly, without personally identifying information.
  • GenAI helped find themes in comments, with staff reviewing and editing the results.
  • The analysis code is public on GitHub, so anyone can see how the findings were reached (though this requires technical skills).

Key takeaway

Transparency about process can matter as much as transparency about data.

Tribal data sovereignty

Tribes are sovereign governments, not stakeholders

“Indigenous experiences under colonial control suggest that data more often means ‘something taken’.”

Desi Rodriguez-Lonebear, “Building a Data Revolution in Indian Country” (2016)

What is Indigenous data sovereignty?

Indigenous data sovereignty

“Indigenous Data Sovereignty recognizes Indigenous Peoples’ inherent rights to access, use, and steward data critical for effective governance, and to govern those data according to their own values, principles, and practices.”1

Blood given by Havasupai Tribe members for diabetes research at Arizona State University was reused for unrelated studies. The university settled in 2010 and returned the samples.

Mismanaged

Cobell v. Salazar: a 15-year class action over the federal government’s management of individual Indian trust accounts and land. Settled for $3.4 billion in 2011.

Miscounted

The 2020 Census undercounted American Indian and Alaska Native people living on reservations by 5.64%, more than any other group.2

“The danger in damage-centered research is that it is a pathologizing approach in which the oppression singularly defines a community.”

“Even when communities are broken and conquered, they are so much more than that—so much more that this incomplete story is an act of aggression.”

Eve Tuck, “Suspending Damage: A Letter to Communities” (2009, 413, 416)

CARE alongside FAIR

FAIR (Findable, Accessible, Interoperable, Reusable) is about making data easy to share and reuse.1 CARE adds people and purpose.2

Collective benefit

“Data ecosystems shall be designed and function in ways that enable Indigenous Peoples to derive benefit from the data.”

Authority to control

“Indigenous Peoples’ rights and interests in Indigenous data must be recognised and their authority to control such data be empowered.”

Responsibility

“Those working with Indigenous data have a responsibility to share how those data are used to support Indigenous Peoples’ self-determination and collective benefit.”

Ethics

“Indigenous Peoples’ rights and wellbeing should be the primary concern at all stages of the data life cycle and across the data ecosystem.”

California context

2011

EO B-10-111

Consultation

Creates the Governor’s Tribal Advisor and directs agencies to consult with California tribes

2017

CalHHS policy2

Agency policies

Departments adopt tribal consultation policies, though most don’t address data

2019

EO N-15-193

Apology

Formal state apology and the Truth and Healing Council

2025

AB 1004; CNRA policy4

Sovereign records

Tribal financial records shared with agencies are exempt from disclosure as “records of a sovereign nation”

Now

CNRA toolkit5

Data sovereignty

Guidance on Indigenous data sovereignty for state partners handling tribal data in natural resource contexts

“Each standard and each category valorizes some point of view and silences another.”

Geoffrey Bowker and Susan Leigh Star, Sorting Things Out (1999)

What this means in practice

Categories aren’t nations

“American Indian and Alaska Native” is a federal race category. It isn’t the same as tribal nations or tribal citizenship, and it isn’t a tribe’s own definition.1

From damage to desire

Tuck’s alternative: “desire-based research frameworks are concerned with understanding complexity, contradiction, and the self-determination of lived lives.”2

Best practices for state entities

  • Consult before collecting or publishing, not after
  • Put data sharing agreements in writing
  • Let tribes review products that identify them
  • Respect a decision not to publish
  • Ask what benefit flows back to the community
  • Document strengths and aspirations, not only harm

If you read one thing from this lecture, make it Eve Tuck’s Suspending Damage: A Letter to Communities.

(Gen)AI governance

When decisions are automated, who’s accountable?

When automation goes wrong

California EDD, 2020–21: under pressure from pandemic-era fraud, the Employment Development Department used a fraud-scoring tool to flag and suspend unemployment claims.

1.1M

claims suspended, with no advance notice to claimants1

600K+

of those were later found to be legitimate

$10B+

paid on fraudulent claims, which drove the crackdown2

Key takeaway

If algorithms aren’t carefully designed, automated decisions can fail at scale, and the cost often falls on the people with the least power to contest them.

The New Jim Code: “the employment of new technologies that reflect and reproduce existing inequities but that are promoted and perceived as more objective or progressive than the discriminatory systems of a previous era.”

Ruha Benjamin, Race After Technology (2019, 6)

How California governs AI

2023

EO N-12-23; AB 3021

Foundations

GenAI executive order; inventory of high-risk automated decision systems

2024

Guidelines; SB 8962

Rules of the road

Procurement guidelines, GenAI risk assessments, and disclaimers plus a path to reach a human

2025

State policy; SB 533

Policy, then regulation

GenAI policy in the State Administrative Manual; frontier developer transparency (the state as regulator)

2026

EO N-5-264

Trusted procurement

Vendor attestations on bias, civil rights, and privacy

ODI in practice: AskCA

AskCA is an AI assistant that helps Californians find state services, built by ODI and launched as a prototype on October 1, 2026.1

  • Chats are kept for 60 days to improve answers, then deleted or anonymized.
  • Conversations aren’t used to train the underlying model.
  • Users are told not to enter Social Security numbers or health information.
  • Collected information may be a public record, subject to exemptions.
  • Users are notified of privacy and data collection practices.

Want to sign up for the beta release? Go to ask.ca.gov!

Questions AskCA raises that the beta release is investigating

Accuracy

“This tool may not always get things right.” What’s an acceptable error rate for this kind of tool?

AskCA: check the source links; accuracy isn’t guaranteed. About · Conditions of use

A path to a human

SB 896 requires a disclaimer and a way to reach a person. What does a good path look like?

AskCA: email and in-chat feedback, no live human support. Conditions of use · About

Public records

Are GenAI chat logs public records? How do you balance transparency with users’ privacy?

AskCA: collected information is a public record, with exemptions. Privacy policy · Notice on collection

Minimization

What is the right retention period? Who decides, and based on what considerations?

AskCA: chats kept 60 days. Privacy policy

Vendor transparency

Which model and vendors sit behind a public service, and where should the public learn that?

AskCA: “third-party service providers.” Notice on collection

Accountability

If a GenAI answer leads someone to miss a deadline, who’s responsible?

AskCA: ODI disclaims liability for errors and omissions. Conditions of use

Putting it together

How ODI approaches this work

Open by default

Publish what the public needs for government transparency and trust.

Minimize by design

Collect and keep only what the purpose requires.

Center people

Start from the needs of the people the data describes and serves.

Govern (Gen)AI

Assess risk, disclose use, test before deploying, and keep humans in the loop.

Questions to ask of any public dataset

  • What is the purpose?
  • Is it necessary?
  • What harms are possible?
  • Who decides?
  • Who is accountable?
  • Who is missing?

Key takeaway

The suppression threshold you’ll pick in a few minutes is a value choice. As Langdon Winner put it, “the things we call ‘technologies’ are ways of building order in our world.”1

Exercise

To publish or not to publish · 30 minutes

The exercise

Part 1

Explore

Load the table and find the smallest cells

Part 2

Suppress

Apply a small-cell rule and see who disappears

Part 3

Recover

Rebuild hidden cells from totals, then fix it

Part 4

Re-identify a fictional person by joining two datasets

Part 5

Decide

Write a one-paragraph recommendation to your director

The small-cell rule

Hide any count of 1 to 10 people, so no one can be singled out from a tiny group. Zeros are shown. This follows California’s health and human services de-identification guidelines.1

Complementary suppression, step by step

0–1718–3435–6465+Total

1. Hide the small cell291917?75The 65+ count (10) is under 11, so it’s hidden.

2. Someone does the math291917107575 − 29 − 19 − 17 = 10. The total gives it away.

3. Hide one more cell2919??75The two add up to 27, but is it 17 + 10? 20 + 7? 26 + 1? No way to tell.

The rule in the exercise

Keep hiding the next-smallest cell in the row until at least two cells are hidden and they add up to 11 or more.

Get set up

Work on your own or in pairs.

git clone https://github.com/lucy-gov/data-ethics-lecture.git
cd data-ethics-lecture
uv sync
uv run jupyter lab exercise/

Data note

All data here is synthetic. County names are real and county sizes are roughly realistic, but every count and every person is made up. No real program, community, or tribe is represented.

Debrief

What you found

28 → 0

hidden cells could be worked out from the published totals, until complementary suppression closed the gap

15% → 29%

of AIAN enrollees in small counties were hidden, before and after complementary suppression

17 of 372

“de-identified” records were matched to a name using only ZIP code, birth year, and sex

Key takeaway

Each fix for privacy had a “cost,” and the cost landed on the same small communities.

Concept: small-cell suppression

Alpine County, White enrollees (synthetic data from the exercise):

0–17 18–34 35–64 65+ Total
True count 29 19 17 10 75
Published 29 19 17 hidden 75

Small-cell suppression

Hide any count based on fewer than 11 people, so no one can be singled out from a tiny group. Zeros are usually shown.

Concept: working backward

The row total is still published. One hidden cell plus a total is one equation with one unknown:

75 − 29 − 19 − 17 = 10

With complementary suppression, one more cell is hidden:

0–17 18–34 35–64 65+ Total
Published 29 19 hidden hidden 75

Complementary suppression

Hide extra cells so the hidden ones can’t be recovered from totals. Now 35–64 + 65+ = 27, but there’s no way to tell how it splits. The cost: a cell of 17 people disappears too.

Concept: linking datasets

“De-identified” extract (no names)

record ZIP born sex benefit
R0279 96120 1957 F $350

Public roster (names, no benefits)

name ZIP born sex
Max Lopez 96120 1957 M
Ruth Calloway 96120 1957 F

A join lines up rows from two tables that share values in the same columns. ZIP code and birth year narrow 3,000+ residents to two people; sex narrows it to one.

Quasi-identifiers

Ordinary details like ZIP code, birth year, and sex that don’t name anyone alone, but often do in combination. Half the records in the extract were unique on just these three.

Let’s discuss

Trade-offs

Suppression, aggregation, or not publishing: what does each cost, and who pays?

Who decides

Whom should the department have consulted about the AIAN rows, and when? (“AIAN” is a race category, not a tribal nation.)

Damage vs. desire

In this exercise, AIAN enrollees showed up only as a privacy risk. Was your memo damage-centered? What would a desire-based release look like?

Linkage

What does “de-identified” actually promise? What should the department have removed before sharing the extract?

Has your answer changed?

A. Publish as is. The Public Records Act presumes access, and these are counts, not individual records.

B. Suppress small cells. Hide counts under 11, plus any cells that would let someone recover them, then publish.

C. Aggregate. Combine small counties into regions, or small groups into broader categories, so no cell is small.

D. Consult first. Hold the release until affected tribes have been consulted, then decide.

E. Withhold. Deny the request and keep the table off data.ca.gov, arguing that the privacy risk clearly outweighs the public interest in disclosure.

Go to menti.com and enter code 9406 1118

QR code linking to the Mentimeter poll

My take: two decisions, not one

Decision 1, the records request: B

  • The Public Records Act is an obligation: release the parts of a record that can be separated out, and justify anything withheld.1
  • Suppress small and complementary cells, cite privacy for those cells only, and release the rest on time.
  • Never share record-level data with quasi-identifiers like ZIP, birth year, and sex intact.

Decision 2, posting on data.ca.gov: D, then maybe C

  • Publication is discretionary, so this is where consultation belongs.
  • Ask affected tribes how the data should look. The answer may be aggregation, an agreed suppression approach, or data-sharing agreements instead of posting.
  • Working out who to consult about “AIAN” rows is itself the classification problem.

The real lesson

The decision is already late. These questions belong at collection and publication planning, with tribes at the table, not on a records-request deadline.

References

Benjamin, Ruha. 2019. Race After Technology: Abolitionist Tools for the New Jim Code. Polity.
Bowker, Geoffrey C., and Susan Leigh Star. 1999. Sorting Things Out: Classification and Its Consequences. MIT Press. https://doi.org/10.7551/mitpress/6352.001.0001.
Brown, Edmund G., Jr. 2011. Executive Order B-10-11. https://www.archive.gov.ca.gov/archive/gov39/2011/09/19/news17223/index.html.
California Department of Justice. 2026. California Consumer Privacy Act (CCPA). https://oag.ca.gov/privacy/ccpa.
California Department of Technology. 2019. Technology Letter 19-01: California Open Data Policy. https://cdt.ca.gov/wp-content/uploads/2019/03/TL-19-01.pdf.
California Department of Technology. 2024. Technology Letter 24-01: GenAI Inventory, Procurement, and Risk Assessment for Executive Order N-12-23. https://www.cdt.ca.gov/policy/technology-letters/technology-letter-24-01/.
California Department of Technology. 2025. Technology Letter 25-01: IT Policies for Generative Artificial Intelligence (GenAI). https://www.cdt.ca.gov/policy/technology-letters/technology-letter-25-01/.
California Department of Technology. 2026. Data.CA.gov: California Open Data Portal. https://data.ca.gov/.
California Government Operations Agency. 2026. Guidelines – California Open Data Handbook. https://handbook.data.ca.gov/guidelines/.
California Government Operations Agency, California Department of Technology, Department of General Services, Office of Data and Innovation, and California Department of Human Resources. 2024. State of California GenAI Guidelines for Public Sector Procurement, Uses and Training. https://cdt.ca.gov/wp-content/uploads/2024/07/3a-GenAI-Guidelines.pdf.
California Health and Human Services Agency. 2017. CalHHS Tribal Consultation Policy. https://www.chhs.ca.gov/wp-content/uploads/2018/11/CHHS-Tribal-Consultation-Policy_1.11.17_FINAL_PDF.pdf.
California Health and Human Services Agency. 2025. Data de-Identification Guidelines (DDG) Template, Edition 2.0. https://www.cdii.ca.gov/wp-content/uploads/2025/12/CalHHS-DDG-Template_Edition2.0.pdf.
California Legislature. 1977. Information Practices Act of 1977, Cal. Civ. Code §§ 1798–1798.78. https://leginfo.legislature.ca.gov/faces/codes_displaySection.xhtml?lawCode=CIV&sectionNum=1798.1.
California Legislature. 2017. AB 90 (Weber), Criminal Gangs, Stats. 2017, Ch. 695. https://leginfo.legislature.ca.gov/faces/billNavClient.xhtml?bill_id=201720180AB90.
California Legislature. 2021. California Public Records Act, Cal. Gov. Code § 7920.000 Et Seq. https://leginfo.legislature.ca.gov/faces/codes_displaySection.xhtml?lawCode=GOV&sectionNum=7921.000.
California Legislature. 2023. AB 302 (Ward), Department of Technology: High-Risk Automated Decision Systems: Inventory, Stats. 2023, Ch. 800. https://leginfo.legislature.ca.gov/faces/billNavClient.xhtml?bill_id=202320240AB302.
California Legislature. 2024. SB 896 (Dodd), Generative Artificial Intelligence Accountability Act, Stats. 2024, Ch. 928. https://leginfo.legislature.ca.gov/faces/billNavClient.xhtml?bill_id=202320240SB896.
California Legislature. 2025a. AB 1004 (Wallis), Tribal Financial Information: Public Records: Exemption, Stats. 2025, Ch. 132. https://leginfo.legislature.ca.gov/faces/codes_displaySection.xhtml?lawCode=GOV&sectionNum=8450.
California Legislature. 2025b. SB 53 (Wiener), Transparency in Frontier Artificial Intelligence Act, Stats. 2025, Ch. 138. https://www.gov.ca.gov/2025/09/29/governor-newsom-signs-sb-53-advancing-californias-world-leading-artificial-intelligence-industry/.
California Natural Resources Agency. 2025. Tribal Consultation Policy. https://resources.ca.gov/-/media/CNRA-Website/Files/Initiatives/Tribal-Affairs/CNRA-Tribal-Consultation-Policy-09252025.pdf.
California Natural Resources Agency. 2026. Indigenous Data Sovereignty for Tribal Stewardship. https://resources.ca.gov/Initiatives/Tribalaffairs/TribalStewardshipPolicyAndToolkit/Explore-the-Toolkit/Indigenous-Data-Sovereignty-for-Tribal-Stewardship.
California Office of Data and Innovation. 2026a. About AskCA. https://ask.ca.gov/about.
California Office of Data and Innovation. 2026b. AskCA Conditions of Use. https://ask.ca.gov/conditions-of-use.
California Office of Data and Innovation. 2026c. AskCA Privacy Notice on Collection. https://ask.ca.gov/privacy-notice-on-collections.
California Office of Data and Innovation. 2026d. AskCA Privacy Policy. https://ask.ca.gov/privacy-policy.
California Office of Data and Innovation. 2026e. California Statewide Data Strategy 2026–2027. https://innovation.ca.gov/reports/california-statewide-data-strategy-2026-2027/.
California Office of Data and Innovation. 2026f. Data Minimization 101. https://hub.innovation.ca.gov/data/minimization-toolkit/101/.
California Privacy Protection Agency. 2026. About Us. https://cppa.ca.gov/about_us/index.html.
California State Auditor. 2016. The CalGang Criminal Intelligence System: As the Result of Its Weak Oversight Structure, It Contains Questionable Information That May Violate Individuals’ Privacy Rights (Report 2015-130). https://information.auditor.ca.gov/reports/2015-130/summary.html.
California State Auditor. 2021. Significant Weaknesses in EDD’s Approach to Fraud Prevention Have Led to Billions of Dollars in Improper Benefit Payments (Report 2020-628.2). https://information.auditor.ca.gov/reports/2020-628.2/summary.html.
Carroll, Stephanie Russo, Ibrahim Garba, Oscar L. Figueroa-Rodríguez, et al. 2020. “The CARE Principles for Indigenous Data Governance.” Data Science Journal 19 (1): 43. https://doi.org/10.5334/dsj-2020-043.
D’Ignazio, Catherine, and Lauren F. Klein. 2020. Data Feminism. MIT Press. https://doi.org/10.7551/mitpress/11805.001.0001.
Engaged California. 2025. Agenda-Setting Data Insights and Methodology: LA Fires Recovery. https://engaged.ca.gov/lafires-recovery/agenda-setting-data-insights/.
Engaged California. 2026a. About the Program. https://engaged.ca.gov/about/.
Engaged California. 2026b. Impact of AI on Work. https://engaged.ca.gov/ai-impact/.
Global Indigenous Data Alliance. 2019. CARE Principles for Indigenous Data Governance. https://www.gida-global.org/careprinciples.
Golle, Philippe. 2006. “Revisiting the Uniqueness of Simple Demographics in the US Population.” Proceedings of the 5th ACM Workshop on Privacy in Electronic Society, 77–80. https://doi.org/10.1145/1179601.1179615.
Legislative Analyst’s Office. 2022. The 2022-23 Budget: Assessing Proposals to Address Unemployment Insurance Fraud. https://lao.ca.gov/Publications/Report/4542.
Madison, James. 1788. The Federalist No. 51. https://guides.loc.gov/federalist-papers/text-51-60.
Mello, Michelle M., and Leslie E. Wolf. 2010. “The Havasupai Indian Tribe Case: Lessons for Research Involving Stored Biologic Samples.” New England Journal of Medicine 363 (3): 204–7. https://doi.org/10.1056/NEJMp1005203.
Moscone, George, and Kenneth Cory. 1972. Argument in Favor of Proposition 11 (Right of Privacy). https://repository.uclawsf.edu/ca_ballot_props/762.
Newsom, Gavin. 2019. Executive Order N-15-19. https://tribalaffairs.ca.gov/wp-content/uploads/sites/10/2020/02/Executive-Order-N-15-19.pdf.
Newsom, Gavin. 2023. Executive Order N-12-23. https://www.gov.ca.gov/wp-content/uploads/2023/09/AI-EO-No.12-_-GGN-Signed.pdf.
Newsom, Gavin. 2026. Executive Order N-5-26 (Trusted AI Procurement). https://www.gov.ca.gov/wp-content/uploads/2026/03/3.30-FINAL-Trusted-AI-Procurement-EO-N-5-26.pdf.
Office of Governor Gavin Newsom. 2025. California to Launch First-in-the-Nation Digital Democracy Effort to Improve Public Engagement. https://www.gov.ca.gov/2025/02/23/california-to-launch-first-in-the-nation-digital-democracy-effort-to-improve-public-engagement/.
Office of Governor Gavin Newsom. 2026. Government, Made Easier: Governor Newsom Introduces AskCA, a New AI-Powered Tool for Californians. https://www.gov.ca.gov/2026/09/09/government-made-easier-governor-newsom-introduces-askca-a-new-ai-powered-tool-for-californians/.
Rodriguez-Lonebear, Desi. 2016. “Building a Data Revolution in Indian Country.” In Indigenous Data Sovereignty: Toward an Agenda, edited by Tahu Kukutai and John Taylor. ANU Press. https://doi.org/10.22459/CAEPR38.11.2016.14.
Sweeney, Latanya. 2000. Simple Demographics Often Identify People Uniquely. Data Privacy Working Paper No. 3. Carnegie Mellon University. https://dataprivacylab.org/projects/identifiability/paper1.pdf.
Tuck, Eve. 2009. “Suspending Damage: A Letter to Communities.” Harvard Educational Review 79 (3): 409–28. https://doi.org/10.17763/haer.79.3.n0016675661t3n15.
United States Indigenous Data Sovereignty Network. 2024. U.s. Indigenous Data Sovereignty & Governance Summit. https://usindigenousdatanetwork.org/us-indigenous-data-sovereignty-governance-summit/.
U.S. Census Bureau. 2022. Census Bureau Releases Estimates of Undercount and Overcount in the 2020 Census. https://www.census.gov/newsroom/press-releases/2022/2020-census-estimates-of-undercount-and-overcount.html.
U.S. Department of the Interior. 2011. Secretary Salazar and Associate Attorney General Perrelli Applaud Final Approval of Cobell Settlement. https://www.doi.gov/news/pressreleases/Secretary-Salazar-and-Associate-Attorney-General-Perrelli-Applaud-Final-Approval-of-Cobell-Settlement.
Walter, Maggie, and Chris Andersen. 2013. Indigenous Statistics: A Quantitative Research Methodology. Left Coast Press. https://doi.org/10.4324/9781315426570.
Wilkinson, Mark D., Michel Dumontier, IJsbrand Jan Aalbersberg, et al. 2016. “The FAIR Guiding Principles for Scientific Data Management and Stewardship.” Scientific Data 3: 160018. https://doi.org/10.1038/sdata.2016.18.
Winner, Langdon. 1980. “Do Artifacts Have Politics?” Daedalus 109 (1): 121–36. https://www.jstor.org/stable/20024652.

Thank you

Questions?

Lucy Andrews, California Office of Data and Innovation

https://innovation.ca.gov
@CaliforniaODI