Briefing Notes – Ethics of AI Institute Meeting

Date: September 18, 2026
Time: 3:00 PM EST
Chair: Matthew Silk
Attendees: Members of the Ethics of AI Institute

• The Institute published its first white paper on AI and agriculture Sowing the machine, based on our discussion at the last meeting.
• Thank you to Kem-Laurin Lubin for spearheading and compiling the white paper.
• Members expressed interest in continuing the practice of producing white papers from Institute meetings.

Defining the Technology

Semi-autonomous weapons

• Human remains "in the loop.".
• A human authorizes or can override an attack.

Fully autonomous weapons

• Operate without real-time human intervention.
• Can independently identify and engage targets.

Lethal vs. Non-lethal systems

• Lethal systems are designed to inflict physical harm.
• Non-lethal systems may perform surveillance, reconnaissance, or other support functions.

Major Ethical Theme 1: Responsibility and Accountability

Central question: Who is responsible when an autonomous weapon causes harm?

Arugment 1: Traditional Military Responsibility Still Applies

• Military command structures already assign responsibility through chains of command.
• Responsibility lies with those who authorize deployment.
• The fact that AI executes the action does not remove human accountability.

Argument 2: AI Complicates Responsibility

• AI may create plausible deniability
• Diffusion of responsibility
• Blame shifting
• Difficulty assigning accountability

Potentially responsible parties include:

• Political leaders
• Military commanders
• Operators
• Software developers
• AI companies
• Data suppliers

Major Ethical Theme 2: Should Weapons Contain Morality?

Central question: should autonomous weapons be equipped with explicit moral frameworks.

Argument 1: Explicit Moral Rules Are Necessary

• Every autonomous weapon already embodies values.
• If moral assumptions are inevitable, they should be made explicit.
• Transparent ethical constraints are preferable to hidden assumptions.

Argument 2: Moral Programming Is Dangerous

• Moral frameworks are culturally dependent.
• Different countries may program conflicting values.
• Governments could embed discriminatory priorities.
• AI could misidentify targets while believing it is acting morally.
• considerable skepticism about whether human moral reasoning can truly be translated into machine logic.

Major Ethical Theme 3: Human vs. Machine Decision Making

Central question: can AI might make better decisions than humans?

Potential Benefits of AI Decision Making

• Greater consistency
• Reduced emotional reactions
• Ability to process vast numbers of scenarios
• Possibility of reducing civilian casualties

Concerns with AI Decision Making

• Goal-oriented systems may pursue objectives in unintended ways.
• AI may optimize outcomes without understanding human values.
• A machine's practical effectiveness may conflict with ethical considerations.

Major Ethical Theme 4: Autonomous Weapons as a Defensive Tool

Central question: can AI improve defensive capabilities?

Potential Advantages

• Can help smaller states defend against larger powers.
• inexpensive drones have enabled countries with fewer resources to resist militarily stronger opponents.
• May reduce military asymmetries.
• Have already changed modern warfare.

Concerns

• Widespread access could fuel arms races.
• Non-state actors may gain access.
• Lower costs may make war easier to initiate.
• The technology could "gamify" warfare by distancing decision-makers from human consequences.

Major Ethical Theme 5: The "Arms Race" Question

Central question: do governments have a moral obligation to develop AI weapons if adversaries are doing so?

Arguments for Development

• Refusing to develop the technology could leave nations vulnerable.
• Development helps create defensive countermeasures.
• Technological preparedness may act as deterrence.

Arguments Against

• Development itself contributes to proliferation.
• New weapons often normalize future escalation.
• History shows technological races can increase conflict rather than security (e.g., nuclear weapons).

Major Ethical Theme 6: Regulation and Governance

The group broadly agreed that self-regulation by private companies is insufficient.

Key Concerns

• Social media as an example of failed self-regulation.
• Financial incentives may conflict with public safety.
• Military AI development is increasingly tied to large technology firms.

Proposed Solutions

• Dedicated government agencies focused on AI oversight.
• Interdisciplinary regulatory bodies.
• Greater transparency and disclosure requirements.
• Independent oversight mechanisms.
• Stronger democratic accountability.

However, participants also acknowledged concerns about governments themselves being influenced by corporate interests or political agendas.

Institutional Trust

A recurring theme was a growing lack of public trust.

• Can private companies be trusted?
• Can governments be trusted?
• Who should govern powerful AI systems?
• Society is experiencing a broader crisis of trust, making AI governance especially difficult.
• Governance systems themselves may require reform before citizens will have confidence in AI regulation.

Technology, Power, and Geopolitics

The discussion explored how AI intersects with global power structures.

Topics included:

• U.S.-China technological competition
• Military-industrial relationships
• The role of technology companies in defense
• Narratives about an "AI race" as rhetoric or real geopolitical competition

3. Future Activities

Publications

The group discussed creating a future publication based on the autonomous warfare discussion.

Next Meeting

Potential Future Topics

• AGI and existential risk ("AI is going to kill us all")
• AI-generated intimate images and related harms
• Child safety and AI-generated content
• AI in education
• Cognitive impacts of AI use
• Surveillance technologies (including Flock systems)
• Civic AI literacy and transparency