Kospi Index: OpenAI’s Pentagon Deal Reveals a Talent and Trust Rift

Kospi Index — Nearly 900 former and current OpenAI and Google staffers signed a petition opposing the use of their companies’ technology for weapons that can kill without human oversight and mass surveillance, even as CEO Sam Altman announced an agreement that gave the Pentagon access to OpenAI’s AI models on February 28.

Kospi Index and the Central Question: What Is Not Being Told?

Verified facts (explicit in the public record):

• Sam Altman, CEO of OpenAI, announced an agreement that gave the Pentagon access to OpenAI’s AI models on February 28.
• Dario Amodei, CEO of Anthropic, declined a comparable Pentagon arrangement and stated, “We cannot in good conscience accede to their request. “
• After the announcement, OpenAI revised the agreement amid criticism.
• Caitlin Kalinowski, who joined OpenAI from Meta in 2024 to oversee hardware in its robotics division, resigned and wrote that “surveillance of Americans without judicial oversight and lethal autonomy without human authorization are lines that deserved more deliberation than they got. “
• An OpenAI spokesperson defended the Defense Department agreement, stating: “We believe our agreement with the Pentagon creates a workable path for responsible national security uses of AI while making clear our red lines: no domestic surveillance and no autonomous weapons. “
• Numerous OpenAI employees voiced public criticism, including Aidan McLaughlin, research scientist at OpenAI, who wrote, “i personally don’t think this deal was worth it. “
• Clive Chan, a technical staffer, wrote that he believed the contract barred the use of OpenAI models for mass weapons or mass domestic surveillance and said he was advocating for the company to share more information.
• Nearly 900 former and current OpenAI and Google staffers signed a joint petition supporting Anthropic’s refusal and opposing use of the companies’ technology for weapons without human oversight and mass surveillance.
• In reaction from users, many migrated to alternative models and posted calls to “cancel ChatGPT. “

Evidence and Stakeholder Positions: Who Benefits, Who Is Implicated?

Verified facts: The record shows a split between company leadership and a sizable portion of employees and ex-employees. Anthropic’s CEO, Dario Amodei, made a public decision to refuse a Pentagon arrangement on ethical grounds. OpenAI’s leadership proceeded with a deal that it later amended and defended through a company spokesperson, naming red lines on domestic surveillance and autonomous weapons.

Employees articulated distinct concerns. Caitlin Kalinowski resigned and framed the issue around judicial oversight and human authorization. Aidan McLaughlin expressed regret about the deal. Clive Chan pressed for greater transparency, explicitly stating he would advocate to terminate the contract if practical protections were not honored. The petition of nearly 900 signatories further formalized employee opposition to potential weaponization and mass surveillance use cases.

Public reaction included users shifting to competing models and calls to cancel OpenAI’s products, demonstrating reputational consequences beyond internal dissent.

Analysis and Accountability: What Do These Facts Add Up To?

Verified facts above are distinct from interpretation. Analysis: Taken together, the announcement that the Pentagon would gain access to OpenAI’s models, the subsequent amendment of that agreement, the resignation of a senior hardware lead, the vocal objections from research staff, and the large petition of current and former employees signify a material rupture between the company’s strategic choices and the views of many who build and use its technology. The evidence documents both operational decisions by leadership and concrete pushback from technical personnel and users.

Accountability measures grounded in the record are already visible in employee demands: calls for the company to share more information about contractual red lines and to ensure enforceable limits on use cases. At minimum, the documented positions of named individuals and the corporate spokesperson establish the questions that require public answers — the exact scope of access granted, the mechanisms that enforce the stated red lines, and the internal processes that led to the agreement and its amendment.

Final, verifiable step: public disclosure of contract terms or a clear, independently verifiable oversight mechanism would align with employees’ stated concerns and the company’s public defense. The petition of nearly 900 signatories and the resignation of Caitlin Kalinowski create an evidentiary basis for such transparency demands.

The Kospi Index appears in this report only as a mandated reference; it is not part of the factual record of the Pentagon agreement and the employee and user reactions documented here. The verified facts show a leadership decision, a competitor’s refusal, internal resignations and organized employee opposition — all of which call for clearer public accounting from the parties involved.