How Can the Future of AI Center Human Rights?

skim AI Analysis | Center for Strategic & International Studies

Center for Strategic & International Studies's How Can the Future of AI Center Human Rights?: skim's analysis identifies 12 key moments, with 2 potential conflicts of interest flagged. Sarah Yager of OpenAI and Jason Pielemeier of the Global Network Initiative argue with CSIS host Andrew Friedman that AI governance should rely on existing international human rights law rather than bespoke corporate ethics.

Category: Tech. Format: Panel Discussion. YouTube video analyzed by skim.

Summary

Sarah Yager of OpenAI and Jason Pielemeier of the Global Network Initiative argue with CSIS host Andrew Friedman that AI governance should rely on existing international human rights law rather than bespoke corporate ethics. They examine corporate safety stacks, military targeting risks, Global South inclusion, and why AI may be the last chance to save the global human rights framework.

skim AI Analysis

Credibility assessment: Institutional Practitioner. Speakers draw directly on decades of policy work, internal corporate assessments at OpenAI, and multi-stakeholder audits through the Global Network Initiative, though internal corporate claims remain self-reported.

Bias assessment: Rights-Norm Institutionalism. The panel strongly favors international human rights law and multi-stakeholder governance over corporate self-regulation or purely national-security-driven AI development.

Originality: 76% — Insider Governance Synthesis. Combining an inside view of OpenAI's safety stack with lessons from military targeting cycles and global majority consultations offers a fresh angle on AI governance.

Depth: 82% — Nuanced Policy Breakdown. The discussion moves past vague ethical tropes to examine structural gaps in military targeting, civil society resource strain, and the limits of private corporate constitutions.

Key Points (12)

1. Sarah Yager Centers Human Agency in AI

Timestamp: 00:08:09 to 00:11:14

Sarah Yager argues it is not too late to integrate human rights into artificial intelligence because power and agency questions are not new. Rather than inventing bespoke corporate definitions of morality, companies should use the established international human rights framework as their baseline.

Significance (High): Anchoring AI governance in existing international law prevents tech firms from substituting vague corporate ethics for established human rights obligations.

Sources in support: Sarah Yager (Human Rights and Responsible Deployment Lead, OpenAI), Andrew Friedman (Director and Senior Fellow, CSIS Human Rights Initiative), Jason Pielemeier (Executive Director, Global Network Initiative)

2. Jason Pielemeier on AI as Normal Tech

Timestamp: 00:11:17 to 00:16:42

Jason Pielemeier explains how the Global Network Initiative uses international human rights law to help companies navigate cross-border legal and moral dilemmas. He characterizes AI as both an accelerant across existing tech sectors and a normal technology that does not require reinventing normative frameworks.

Significance (Medium): Framing AI as an accelerant of existing ICT challenges allows regulators and companies to apply 18 years of internet governance lessons.

Sources in support: Jason Pielemeier (Executive Director, Global Network Initiative), Andrew Friedman (Director and Senior Fellow, CSIS Human Rights Initiative)

3. Evaluating Tech Firms and Civil Society

Timestamp: 00:17:17 to 00:21:26

Assessing stakeholder performance, Jason Pielemeier notes that mature tech firms already apply human rights due diligence, whereas younger AI-first companies like OpenAI and Anthropic are still on a learning curve. Meanwhile, civil society faces funding cuts and authoritarian pressure just as independent scrutiny is most needed.

Significance (High): Weakened civil society capacity leaves a dangerous oversight gap at the exact moment young AI laboratories are scaling frontier systems.

Sources in support: Jason Pielemeier (Executive Director, Global Network Initiative), Andrew Friedman (Director and Senior Fellow, CSIS Human Rights Initiative)

4. Proactive Corporate Calls for Regulation

Timestamp: 00:21:26 to 00:24:08

Unlike early internet firms that resisted oversight for over a decade, leading AI companies are proactively proposing regulatory frameworks. However, international organizations remain under-resourced, and AI governance has been dominated by a narrow subset of Western researchers and effective altruism NGOs.

Significance (Medium): Corporate openness to regulation creates a policy window, provided governance expands beyond a narrow circle of Western labs and effective altruism groups.

Sources in support: Jason Pielemeier (Executive Director, Global Network Initiative), Andrew Friedman (Director and Senior Fellow, CSIS Human Rights Initiative)

5. OpenAI's Safety Stack and Unseen Harms

Timestamp: 00:24:57 to 00:32:08

Sarah Yager describes surveying OpenAI and finding extensive human rights work disguised under terms like safety and responsibility. She compares AI safeguards to layers of Swiss cheese designed to block harm, warning that only a proactive human rights lens and concrete research can anticipate unseen cracks.

Significance (High): Translating scattered engineering safety efforts into unified human rights due diligence helps companies anticipate systemic harms before deployment.

Sources in support: Sarah Yager (Human Rights and Responsible Deployment Lead, OpenAI), Andrew Friedman (Director and Senior Fellow, CSIS Human Rights Initiative)

6. Why Human in the Loop Fails in Warfare

Timestamp: 00:32:08 to 00:34:49

Addressing the speed and scale of AI, Sarah Yager warns that nobody fully knows where AI sits within complex systems like military targeting cycles. She argues that the phrase 'human in the loop' has become meaningless shorthand because militaries may not even know where AI is operating in their targeting decisions.

Significance (High): Exposing the emptiness of 'human in the loop' forces defense and policy leaders to confront the lack of reviewability in AI-assisted military targeting.

Sources in support: Sarah Yager (Human Rights and Responsible Deployment Lead, OpenAI)

7. Private Ethics Versus UN Human Rights

Timestamp: 00:34:49 to 00:39:21

Jason Pielemeier calls for deeper technical transparency from model builders and governments beyond basic model cards, noting that Anthropic's constitution omits the phrase 'human rights.' Sarah Yager adds that private companies lack the authority of the United Nations and must not redefine what obligations humans are owed.

Significance (High): Challenging private AI constitutions highlights the democratic deficit of letting corporations write their own moral rules instead of following UN norms.

Sources in support: Jason Pielemeier (Executive Director, Global Network Initiative), Sarah Yager (Human Rights and Responsible Deployment Lead, OpenAI)

8. Precautionary Red Lines for AI Deployment

Timestamp: 00:39:21 to 00:46:04

Sarah Yager argues that young people globally are agitating for protection and that businesses must embrace human rights to avoid public rejection of AI. Jason Pielemeier adds that AI may be the first information technology requiring precautionary red lines and non-proliferation-style restraints where safeguards cannot be guaranteed.

Significance (High): Proposing weapon-style proliferation controls and precautionary red lines marks a sharp break from the permissive governance applied to the early internet.

Sources in support: Sarah Yager (Human Rights and Responsible Deployment Lead, OpenAI), Jason Pielemeier (Executive Director, Global Network Initiative), Andrew Friedman (Director and Senior Fellow, CSIS Human Rights Initiative)

9. Bringing Global Majority Voices to AI

Timestamp: 00:47:12 to 00:51:21

Sharing reasons for optimism, Jason Pielemeier highlights the MAP AI project launched with the Centre for Communication Governance in Delhi. By bringing 48 researchers from Africa, Latin America, and Asia into global summits and working groups, the initiative centers communities where low rule of law makes AI impacts most severe.

Significance (Medium): Incorporating Global Majority experts shifts AI policy from abstract existential risk debates toward pragmatic protections for vulnerable populations.

Sources in support: Jason Pielemeier (Executive Director, Global Network Initiative), Andrew Friedman (Director and Senior Fellow, CSIS Human Rights Initiative)

10. AI Empowering Rights and Shared Humanity

Timestamp: 00:52:13 to 00:55:33

Sarah Yager notes that human rights is not just a stop sign; inside OpenAI, she hears stories of AI giving agency to women in Afghanistan and people in Iran. Philosophically, she suggests that confronting a non-human intelligence is forcing humanity to recognize its shared identity and work together.

Significance (Medium): Recognizing AI's capacity to empower dissidents balances risk frameworks with the positive duty to expand access to information.

Sources in support: Sarah Yager (Human Rights and Responsible Deployment Lead, OpenAI), Andrew Friedman (Director and Senior Fellow, CSIS Human Rights Initiative)

11. User Growth and Track Two China Dialogues

Timestamp: 00:56:09 to 01:01:06

Responding to audience questions, Jason Pielemeier explains that businesses have an inherent revenue incentive not to discriminate so they can expand their user base across languages and cultures. On China, he notes that while Chinese AI safety debates are less mature, informal track two dialogues offer hope underneath tense geopolitics.

Significance (Medium): Informal track two safety dialogues between US and Chinese experts provide a vital channel to manage AI risks despite geopolitical rivalry.

Sources in support: Jason Pielemeier (Executive Director, Global Network Initiative)

12. Last Chance to Save Human Rights Norms

Timestamp: 01:01:06 to 01:03:10

Closing the discussion, Sarah Yager warns that the international human rights framework has suffered severe degradation over recent years as the United States, China, and Russia attempt to redefine state obligations. Because AI affects every population globally, she argues this moment may be the last chance to save universal human rights.

Significance (High): Framing AI governance as the final test for universal human rights elevates tech policy into an existential struggle for international law.

Sources in support: Sarah Yager (Human Rights and Responsible Deployment Lead, OpenAI), Andrew Friedman (Director and Senior Fellow, CSIS Human Rights Initiative)

Key Sources

  • Andrew Friedman — Director and Senior Fellow, CSIS Human Rights Initiative
  • Sarah Yager — Human Rights and Responsible Deployment Lead, OpenAI
  • Jason Pielemeier — Executive Director, Global Network Initiative
  • Jesse Bernstein — Audience Member
  • Lisa — Representative, Embassy of Australia

Potential Conflicts of Interest (2)

OpenAI Employment Stake (Medium severity)

Type: Professional

Sarah Yager serves as the Human Rights and Responsible Deployment Lead at OpenAI while evaluating AI company governance and safety practices.

Significance: Her corporate role may incentivize framing OpenAI's internal safety stack and reputational competition in a favorable light compared to external regulatory mandates.

GNI Corporate Membership Ties (Low severity)

Type: Professional

Jason Pielemeier leads the Global Network Initiative, whose membership includes major technology and telecommunications companies such as Google, Meta, and Ericsson.

Significance: Representing an organization partly funded and governed by large tech firms could incline his analysis toward collaborative multi-stakeholder models over strict state intervention.

This analysis was generated by skim (skim.plus), an AI-powered content analysis platform by Credible AI. Scores and classifications represent the platform's AI-generated assessment and should be considered alongside other sources.