Tech companies pledge to protect 2024 elections from AI-generated media

At the Munich Security Conference, 20 tech companies (including OpenAI, Microsoft, Amazon, Meta, TikTok, X, Adobe, and others) agreed to an AI elections accord committing to improve detection, labeling, and provenance of synthetic media (e.g., via C2PA-like standards), conduct model risk assessments, and track viral AI-generated content, while acknowledging significant limitations and the likelihood that bad actors will exploit open/jailbroken models. Recent deepfake incidents (e.g., a fake Macron news segment and Biden-voice robocalls) illustrate the threats; firms are exploring watermarking/detection and cautious release practices (e.g., limited access to OpenAI’s Sora), amid ongoing debates balancing free expression with preventing deceptive synthetic political content.