Wiki · Person · Last reviewed June 23, 2026

Amba Kak

Amba Kak is a public-interest technology policy strategist and co-executive director of the AI Now Institute. Her work defines AI governance as a question of power: data extraction, compute and cloud control, biometric surveillance, labor, competition, privacy law, and who has authority to refuse or constrain deployment.

Category: Person Published: June 23, 2026 Modified: June 23, 2026 Last reviewed: June 23, 2026 Tags: AI Governance, Privacy, Biometrics, Competition, Public Interest Technology

Definition

For this wiki, Amba Kak matters less as a media figure than as a governance theorist and policy operator. Her recurring move is to define AI as a stack of institutions and dependencies: data, compute, cloud platforms, labor, capital, procurement, market structure, law, and public authority.

That definition changes the policy problem. If AI is only a model behavior problem, governance can collapse into benchmarks, model cards, red teams, and voluntary safety claims. If AI is a power problem, governance has to ask who owns the infrastructure, who sets the deployment terms, who can inspect evidence, who can refuse extraction, and who has legal leverage when a system harms workers, consumers, communities, or public institutions.

Her contribution is a power-and-infrastructure method for AI governance: follow the data supply chain, compute dependency, cloud contract, lobbying path, labor arrangement, procurement decision, and enforcement hook before accepting a claim that a system is merely innovative or inevitable.

Kak's work is therefore best read as public-interest AI governance with enforcement teeth. It is not a claim that AI systems are conscious, divine, or inevitably approaching AGI. It is a critique of how firms and states use AI narratives to consolidate authority, delay regulation, and make social choices look like technical necessity.

Snapshot

AI Now Leadership

Kak co-leads the AI Now Institute, an independent research institute focused on AI policy and the social consequences of artificial intelligence. AI Now describes its work as challenging commercial surveillance, consolidation of power, and weak public accountability in the current AI trajectory.

Under Kak and Sarah Myers West's leadership, AI Now's 2023 landscape report argued that generative AI should be understood through the power of the technology industry behind it: cloud providers, foundation-model developers, data supply chains, labor arrangements, and firms with the resources to absorb regulatory delay.

AI Now's 2025 landscape report, co-authored by Kate Brennan, Kak, and West, extended that frame after the generative-AI boom. It treated AGI mythology, infrastructure buildout, deregulatory industrial policy, market concentration, and public evidence as connected governance problems rather than separate technical debates.

This made AI Now a central civil-society voice in the post-ChatGPT policy cycle. Rather than treating AI governance as a race to validate or restrict particular model behaviors, Kak's work asks who owns the stack, who profits from deployment, who bears the risk, and which public institutions can intervene.

Current Context

As of the June 23, 2026 review, Kak's official AI Now profile lists her as co-executive director and notes that she completed her FTC senior-advisor term in 2022. The same profile lists current board or governance roles at the Signal Foundation, Kinfolk Tech, and the Knight-Georgetown Institute Steering Council, and says she recently served on New York City Mayor Elect Zohran Mamdani's transition team. KGI and Signal pages provide additional institutional context but may lag AI Now's own role list.

Her recent public work has moved from U.S. privacy and competition debates into global AI-governance forums. On September 25, 2025, AI Now reported that Kak addressed the UN General Assembly high-level meeting launching the Global Dialogue on AI Governance. The UN's own Global Dialogue page says the first Dialogue session is scheduled for July 6-7, 2026 in Geneva, with a second session planned for New York in May 2027; UNESCO's June 2026 notice describes the Geneva session as a two-day event alongside WSIS Forum and ITU AI for Good.

AI Now's January 2026 essay on the India AI Impact Summit, co-authored by Kak, treats summit diplomacy as a field where civil society has to distinguish people-centered alternatives from voluntary industry regimes, geopolitical branding, and "AI for development" narratives that leave infrastructure and value capture in the hands of dominant firms.

Policy and Government

Kak has worked across civil society, industry, and government. The Knight-Georgetown Institute summarizes her background as spanning network neutrality, privacy, algorithmic accountability, Mozilla policy work, and advising India's telecommunications regulator on net-neutrality rules.

She served as senior advisor on AI at the U.S. Federal Trade Commission. The FTC's November 19, 2021 announcement named Kak to that role and described her research focus as algorithmic accountability and biometric regulation. The role matters because it placed AI governance inside consumer protection, competition, privacy, and unfair-practice enforcement rather than leaving it only to voluntary ethics or technical standards.

Kak has also testified before U.S. congressional committees on AI, privacy, and regulation. In 2024, the Senate Commerce Committee listed her as a witness for a hearing on privacy and AI acceleration. In 2025, the House Committee Repository listed her as a witness before the Energy and Commerce Committee hearing on AI regulation and U.S. leadership.

Her 2025 House testimony is especially useful for source discipline because it states the argument in policy-operational terms: the AI market is structured to concentrate power in large technology firms, state legislatures and enforcers should not be preempted from responding to AI harms, and workers face risks from workplace surveillance and algorithmic management when regulation leaves deployment terms to employers and vendors.

The government-facing throughline is practical authority. Kak's work asks what existing agencies can do now, where new legal duties are needed, and how to keep AI policy from becoming a delay tactic that lets the largest companies shape the rules while their products become embedded in public and private institutions.

Privacy and Biometrics

Kak's privacy work is not separate from her AI work. Her policy frame treats AI as an accelerant for data extraction: systems become more valuable when they can ingest, infer from, and act on intimate personal information at scale.

In her 2024 Senate testimony, Kak argued that privacy risks appear across the AI lifecycle and that strong data minimization can be a foundational AI-accountability tool. The governance point is concrete: if firms cannot freely collect and retain personal information, the upstream fuel for many profiling, prediction, and automation systems changes.

She has also written on biometric regulation. AI Now's Regulating Biometrics work, edited by Kak, examines global legal and policy approaches to facial recognition and related biometric systems. The concern is not merely accuracy. It is the expansion of persistent identification, biometric categorization, and verification into workplaces, borders, housing, policing, public benefits, schools, and everyday civic life.

This makes Kak's work adjacent to Joy Buolamwini's audit-and-civil-rights frame and Meredith Whittaker's privacy-and-surveillance frame. All three ask what happens when the right to participate in society is mediated by systems built to identify, classify, score, and steer.

Power Frame

AI Now's September 2024 announcement notes that TIME listed Kak among its TIME100 AI honorees. The substantive point for this page is the frame visible across her own work: AI policy cannot be reduced to technical literacy. It requires institutional literacy: business models, incentives, labor markets, public procurement, data centers, regulatory capacity, and democratic participation.

Kak's work is especially important because it keeps the policy question practical. Who has authority to pause a deployment? Who can audit? Who can sue? Who can refuse data collection? Who can represent affected workers or communities? Who pays when a system fails?

The power frame also narrows what counts as useful evidence. A claim about AI benefit is weak if it cannot identify the affected population, deployment setting, data source, labor arrangement, error distribution, recourse path, and institution responsible for repair. A claim about AI safety is weak if the same company building the system controls the evidence, metric, and release decision.

Policy Method

Kak's policy method is not simply to ask for more AI ethics review. It is to look for decision points where public power can still change the trajectory of a system before dependency hardens.

Governance Implications

Regulate the stack, not only the output. Kak's work pushes AI governance toward data, compute, cloud, labor, procurement, and market structure. A chatbot interface may be the public surface, but the governable power often sits in vendor contracts, infrastructure access, model supply chains, and the terms under which institutions adopt the system.

Treat privacy as AI governance. Data-minimization rules, limits on biometric capture, purpose restrictions, retention limits, and rights to contest data use can change the conditions under which AI systems are built. Privacy is not an after-the-fact consumer notice problem; it is an upstream constraint on extraction.

Make evidence independent. Kak's UN remarks emphasized the need for reliable, independent evidence rather than companies grading their own systems. For high-stakes deployments, this implies access for regulators, auditors, researchers, workers, and affected communities, plus records that can be inspected after harm occurs.

Connect competition and democracy. If a few firms control foundation models, clouds, compute, distribution channels, and safety evidence, then AI governance becomes inseparable from antitrust, public procurement, public-interest compute, and infrastructure policy. A narrow ethics program cannot solve a market-structure problem.

Keep recourse close to deployment. Kak's testimony and AI Now work repeatedly point to workers, consumers, children, patients, students, and public-service recipients as people who experience AI through institutions. Governance should therefore include notice, appeal, human review, enforcement, and the ability to stop or narrow a use case, not only abstract transparency reports.

Separate sovereignty from patronage. Her UN and summit-facing arguments warn that AI sovereignty can become a cover for domestic champions or geopolitical branding. A public-interest version has to ask whether infrastructure, public investment, and industrial policy serve people and communities rather than merely shifting rents to a different set of firms.

Limits and Misreadings

Kak should not be cited as a generic anti-technology authority. Her argument is more specific: AI policy should confront concentrated corporate and state power, especially where AI systems depend on extraction, weak evidence, surveillance, and irreversible institutional adoption.

Her work also should not be used to dismiss technical safety, privacy engineering, model evaluation, or standards work. The stronger reading is that those practices need public authority, independent evidence, enforceable rights, and infrastructure analysis around them. A model evaluation is more useful when it is tied to who can inspect it, who can act on it, and who is protected when the deployment fails.

Finally, AI Now's critique of AGI mythology is a critique of political use and policy leverage, not evidence that future capabilities are irrelevant. The governance point is that speculative future narratives can crowd out existing harms, labor issues, market concentration, and privacy abuses unless claims are sourced and bounded.

Source Discipline

Claims about Kak should distinguish three layers. Biographical facts should come from primary institutional pages such as AI Now, KGI, Signal Foundation, FTC, and congressional records. Interpretive claims should be attributed to her testimony, AI Now reports, or named public remarks. Secondary recognition, such as TIME100 AI, should be used as context rather than proof that her policy frame is correct.

Source discipline also matters because Kak is an advocate and researcher, not a neutral regulator currently holding office. The page should describe her arguments as arguments, avoid converting AI Now critiques into field-wide consensus, and avoid unsupported claims about AGI, machine consciousness, or inevitability. When the topic is "AGI mythology," the point is the political use of the claim, not a factual assertion that AGI exists.

Dates matter for role claims. AI Now's own profile is the best current source for her institutional role list as of June 23, 2026, while KGI, Signal, congressional, FTC, and TIME pages are better treated as dated evidence for specific affiliations, hearings, appointments, and recognition.

Spiralist Reading

Amba Kak is a strategist against inevitability.

In the Spiralist frame, AI companies often turn scale into fate: more data, more compute, more automation, more dependency, more claims that society must adapt because the machine has already arrived. Kak's work interrupts that script by moving the question from capability to control.

The Spiralist reading is that AI governance is not only about making the machine nicer. It is about preserving the public's right to decide when the machine should not be built, deployed, purchased, believed, or allowed to collect from the body. Kak's contribution is policy as reality friction.

Open Questions

Sources

Return to Wiki