OpenAI deletes ban on using ChatGPT for "military and warfare"

OpenAI’s quiet removal of a blanket ban on “military and warfare” uses of ChatGPT has reignited debate over the role of commercial AI in armed conflict. Commenters weigh whether refusing military work is a meaningful ethical stance or merely cedes influence to less constrained actors, citing real-world uses of AI for targeting, drones, and information operations. Many see the change as a predictable response to defense-sector money and strategic pressure, while others worry it accelerates an AI‑driven arms race with unclear safeguards and accountability.

Scope of the Policy Change

  • Old policy explicitly banned “military and warfare.”
  • New policy drops that phrase but still bans “develop or use weapons” and “use our service to harm yourself or others.”
  • Some see this as a narrow language normalization: focus on harmful activities rather than specific actors like militaries.
  • Others see it as a major ethical retreat that opens the door to broad military work while only excluding direct weapons design.

Motivations & Governance

  • Many commenters attribute the change to money and access to huge defense budgets, tying it to OpenAI’s commercialization trajectory and board/Altman turmoil.
  • Several argue “unilateral disarmament” by one company is pointless; if OpenAI abstains, others (or nation-states) will fill the gap.
  • A minority speculate the explicit change might function as a kind of “warrant canary” for growing defense involvement.

Existing and Anticipated Military Uses of AI

  • Cited examples: AI-assisted target selection in Gaza, war-gaming for Ukraine’s counteroffensive, and pervasive drone use (including FPV and emerging computer-vision-guided drones).
  • Others highlight mundane but large-scale uses: logistics, database work, report writing, declassification, and disaster planning (e.g., Corps of Engineers).
  • Information warfare is a major concern: LLM-powered propaganda, astroturfing, and AI-written news; some believe this is already happening.

Ethical Arguments & Collective Responsibility

  • One camp: it’s more ethical to participate and shape “safer” systems than to leave the field to adversaries; analogies drawn to nukes and deterrence.
  • Opposing view: “everyone else is doing it” is a weak moral defense; AI involvement could intensify harm, bias, and civilian casualties.
  • Some compare this moment to the Manhattan Project and question the responsibility of employees, users, and even data contributors.

Broader Reactions

  • Many see the move as part of a long pattern of tech firms abandoning aspirational ethics (“don’t be evil,” “for the good of humanity”) once growth and state power align.
  • Skeptics also question relying on current LLMs for life-and-death decisions given hallucinations and bias.