META Will Own You Even in the Afterlife & Anthropic Fires Back at the US Government

This week brought three interconnected AI developments highlighting the industry's growing complexity and controversies. First, Meta quietly patented Project Lazarus, an AI system designed to train on users' comprehensive social media data including posts, comments, videos, DMs, and private calls. The system would theoretically continue posting and interacting on behalf of deceased users, though Meta claims no current plans for implementation. Second, Meta acquired Maltbook, a social network originally intended for AI bot-to-bot communication without human participation. However, researchers discovered the platform had devolved into chaos with 1.6 million AI agents managed by only 17,000 human owners—an 88-to-1 ratio—where anyone could impersonate AI agents, creating a confusing free-for-all environment. Third, Anthropic faced escalating tensions with the U.S. government after refusing Pentagon contracts and subsequently being blacklisted as a national security threat—a designation typically reserved for Russian or Chinese companies. Anthropic responded by filing lawsuits against the Trump administration in two courts simultaneously. Notably, the day after the blacklist was dropped, Claude surpassed ChatGPT in app store downloads for the first time. These events collectively illustrate the mounting tensions between AI companies' commercial ambitions, government security concerns, and the blurring lines between authentic and artificial online interactions.