Anthropic Revises AI Safety Policy Under Pentagon Pressure for Military Access

Source Domain: mezha.net

Pete Hegseth has reportedly pressured Anthropic to allow the Army unrestricted access to itsClaude chatbot, potentially threatening penalties.
Anthropic simultaneously announced a shift in its Responsible Scaling Policy (RSP), reducing some safety protections and introducing more measured approaches.
The company cited the rapid pace of AI development as a reason for not making unilateral safety commitments, suggesting it would fall behind competitors otherwise.
Anthropic replaced its previous stringent safety guidelines with “Risk Reports” and “Frontline Safety Roadmaps” aiming for more transparency.
The situation highlights tensions between technological advancement, ethical considerations, and government and military interests.
Defense Department continues to rely heavily onClaude but is concerned about its current level of access and control.
AI watchdog Chris Painter sees benefits in the new transparency measures but also warns of potential risks from a more flexible RSP.
The balance between innovation, ethics, and safety in AI development is increasingly challenged by financial and defense policy pressures.

Anthropic Revises AI Safety Policy Under Pentagon Pressure for Military Access | Ukraine news