TL;DR: A former Google DeepMind engineer has proposed a detailed governance framework for AI contracts with government entities, focusing on human control and data privacy.
Summary: A former Google DeepMind engineer has introduced a governance framework for AI provided to government entities with coercive powers. It mandates human control in targeting and force decisions (excluding defensive systems) and prohibits AI profiling of individuals not under specific investigation. The framework also establishes a seven-person Defense AI Review Agency to assess compliance and publish annual transparency reports.
Why it matters: This signals growing pressure for ethical AI deployment in public sectors, potentially influencing future regulations and procurement standards. AI developers should anticipate increased scrutiny on human-in-the-loop mechanisms and data privacy in government contracts.
Source: rss