What safeguards did OpenAI agree to with the Pentagon?

Question

Accepted Answer

OpenAI Agreed to Specific Safeguards with the Pentagon

OpenAI has entered into an agreement with the Pentagon that incorporates specific safeguards, mirroring restrictions previously proposed by AI rival Anthropic. These safeguards are designed to govern the use of OpenAI's models within the US defense agency's classified network.
Prohibition on Domestic Mass Surveillance

A core tenet of the agreement between OpenAI and the Pentagon is the explicit prohibition of domestic mass surveillance. This principle is integrated into OpenAI's safety guidelines, which are part of the formal deal. The company's stance, as articulated by CEO Sam Altman, emphasizes that its technology should not be employed for such purposes.
Human Responsibility for Use of Force

The agreement also mandates human responsibility for the use of force, including in scenarios involving autonomous weapon systems. This safeguard ensures that critical decisions related to lethal force remain under human oversight. OpenAI CEO Sam Altman has publicly stated that humans must remain involved in high-stakes automated decision-making processes.
Similar Red Lines to Anthropic

The safeguards included in OpenAI's deal with the Pentagon are described as having "similar red lines" to those proposed by Anthropic. This suggests a shared commitment to ethical boundaries regarding military applications of AI. Anthropic had also insisted on similar principles, refusing unconditional military use of its Claude models due to concerns about mass surveillance and autonomous weapons.

Reading history

Hide →

This history is cleared when the page is closed.

What safeguards did OpenAI agree to with the Pentagon?

Loading

Was this article helpful?

Helpful

Not helpful

Report an issue

1

0

0

Share with friends