Crash Override LLC has launched PNN, a live satirical news channel that deploys five fictional AI anchors to react to real-time headlines and market data. Unlike traditional automated news bots, PNN emphasizes persistent character continuity, with anchors maintaining moods, running feuds, and memories across segments. The project, which surfaced on Hacker News on October 2, 2026, represents a shift from static text-to-speech outputs to dynamic, context-aware AI performance.

Architecture and Data Sources

The platform operates without pre-scripted segments, relying on AI to generate content moments before broadcast. While headlines are sourced from real outlets credited on screen, the anchors' reactions and jokes are entirely synthetic. This approach requires robust real-time data ingestion and low-latency generation pipelines to maintain the illusion of live broadcasting. The system explicitly warns that AI errors are possible, urging users to verify facts with original sources before sharing.

Voice Synthesis and Legal Framework

PNN utilizes open speech datasets for its audio output, specifically LibriTTS-R (Koizumi et al., CC BY 4.0) and VCTK (University of Edinburgh, CC BY 4.0). These datasets have been modified to ensure no anchor voice is a direct copy of any single human speaker, mitigating potential likeness rights issues. The project also features 'parody callers,' AI-generated stand-ins for public figures, who are explicitly labeled as fictional and not affiliated with the real individuals they mimic.

Key Takeaways

  • PNN uses open-source TTS models (LibriTTS-R, VCTK) with modified voices to avoid direct impersonation.
  • The system maintains stateful memory for characters, allowing for recurring jokes and feuds.
  • Content is live and unscripted, reacting to real-time news and market data feeds.
  • The project is operated by Crash Override LLC and is strictly satirical, with no financial or legal advice provided.

The Bottom Line

PNN demonstrates how persistent state and open-source TTS can create more engaging, albeit fictional, live media experiences without requiring human actors.