X is close to completing a full rewrite of its legacy search code, according to Nikita Bier, the platform’s head of product. The announcement came via a post on February 21, 2026, confirming what many users had already suspected — that X search had become increasingly unreliable. Bier attributed the degradation to AI agents hammering the system and choking it at scale, pushing the small engineering team to rebuild the feature from the ground up alongside improved bot detection measures.
AI Agents Are Breaking Social Platforms
The problem extends beyond X. As AI tools become more widely available, platforms like Meta are facing the same wave of automated, low-quality content flooding their systems. Bier himself acknowledged there is no “silver bullet” solution, warning that the proliferation of AI tools will create headaches across all major social apps. The real danger lies in what this AI-generated content does to the data pipelines that power AI models in return — for X, that means xAI’s Grok risks being trained on increasingly synthetic, low-value posts.
A Conflict at the Heart of Big Tech’s AI Push
There is a striking contradiction in how platforms are responding. X and Meta are simultaneously spending hundreds of billions on AI development while scrambling to limit AI misuse on their own platforms. Promoting AI-generated content creation tools while trying to suppress AI spam puts both companies in a difficult position — one that could frustrate users and undermine the very models they are investing so heavily to build.
For now, X’s search rewrite represents a necessary step in keeping the platform functional as automated activity continues to surge.