Living Neurons Play Doom Better Than Your Intern

Living Neurons Play Doom Better Than Your Intern

Cortical Labs connected 200,000 lab-grown human neurons to an LLM, with neuron firing patterns directly influencing token generation in real-time gameplay [4][5][6]. Their biocomputer learned Doom in under a week, advancing from their 2022 Pong demonstration that used 800,000 cells.

This isn't a gimmick — it's the first practical demonstration of biological-digital hybrid intelligence where living tissue makes computational decisions. The neurons receive electrical feedback and adapt their firing patterns, essentially learning through biological processes while interfacing with silicon systems.

The ethical implications are staggering. We're using human-derived cells as computational components, raising questions about consciousness, consent, and the nature of intelligence itself. If these systems scale, we're looking at a future where the line between biological and artificial intelligence disappears entirely.

OpenAI Adds Surveillance Guardrails to Pentagon Deal

Sam Altman announced OpenAI is amending its Department of Defense contract to explicitly prohibit domestic surveillance of US persons and bar intelligence agencies like the NSA from using their systems [7][8][9]. The changes come after intense backlash over potential mass surveillance applications and autonomous weapons development.

The amendments represent damage control more than principle. OpenAI faced criticism for abandoning its original military restrictions while competitors like Anthropic maintained stricter ethical boundaries. The company is trying to thread the needle between lucrative government contracts and maintaining public trust.

This sets a precedent for AI companies navigating military partnerships. The explicit surveillance prohibitions may become standard contract language, but enforcement remains questionable. When national security interests clash with corporate promises, history suggests the promises bend first.