New Manmade Horrors Just Dropped...

Storyline / Premise

• The video centers on “man-made horrors” created by mapping animal brains—especially a male fruit fly’s brain (~166,000 neurons)—into computer simulations and then embedding that brain activity into games.
• The creator, Mutahar, frames it as creepy because the brain is real (based on scientific mapping), but it’s not presented as a truly sentient experience of a living fly.
• Instead, the project is more like a biologically inspired controller puppeted inside virtual worlds.
• He explores multiple projects that demonstrate this concept, moving from a “fly escape room” concept to Minecraft mob flies, then into a Doom shooter environment where the fly-brain is connected to gameplay.

Gameplay / Project Highlights (and key observations)

Fly Escape Room (NPCs powered by fruit fly brain)

• A virtual environment is populated with objects (food and household items like apples/bananas/vinegar/dirty dishes/a fan).
• Flies are spawned into the simulation; their behavior is driven by the mapped brain model.
• Observed behavior includes:
  • Hitting walls
  • Interacting in odd ways with objects
  • “Escape” style reactions
• The creator emphasizes the uncanny factor: it feels like watching life inside a simulation, even though the intent is emulation/experimentation.

Running fly brain locally (“Flybrain” project)

• Mutahar demonstrates how to run a fly brain simulation on his computer using a Python setup with GPU acceleration.
• A UI/monitor shows brain components like retina activation and sensory-related inputs.
• Notes on performance and behavior:
  • The simulation runs slowly/roughly (he mentions low FPS, e.g., ~3 FPS)
  • It doesn’t behave exactly as expected
• He notes that different sensory/stimulus inputs (examples mentioned: sugar, looming, grooming, bitter, olfactory) can change neural firing patterns.

Fruit fly brain inside Minecraft (“fly brain Minecraft”)

• Using a GitHub project, he injects a fruit fly brain as a Minecraft mob.
• He spawns male fruit fly(s) using commands like /fruit fly spawn.
• Behavior:
  • The fly tends to spin and flee when the player gets near
  • He suspects limitations in the harness/implementation cause it to remain mostly in constrained spinning/escape behavior
• Performance:
  • Spawning multiple flies increases CPU load heavily (he observes a noticeable spike)

Fruit fly brain in Doom (“Doom Fly” / Doom Fly project)

• The creator launches a Doom-integrated setup where the fly brain drives actions in the Doom environment.
• Notes:
  • CPU usage spikes (he cites around ~80%)
  • The fly “looks around” and fires at enemies
  • He highlights neural activity/sensory input in a web interface
• He suggests the fly dies quickly, implying it’s not built to reliably “win” yet and may be limited by performance/acceleration.

OpenWorm (context / comparison)

• He brings up OpenWorm, an open-source project dating back to around 2011, which allows people to run an emulated worm nervous system in containers.
• He uses it as a comparison point: the worm is smaller than the fly brain, but it still demonstrates “life-like” biological control inside software.

Strategies / Key Tips Mentioned

• No formal, step-by-step “strategies” for winning were provided.
• The practical “how-to” elements were mainly about running and spawning the simulations:
  • Use a local environment (Python + GPU) to run the fly simulation.
  • In Minecraft:
    • Start a world (single player, creative mentioned).
    • Spawn mobs via command, e.g., /fruit fly spawn
      • He notes it’s “fruit fly spawn”, not “fly brain spawn”.

Ethics / Commentary (important themes)

• Mutahar repeatedly highlights the disturbing feel of:
  • Mapping a real brain into game worlds
  • Then subjecting it to hostile or “torture-like” gameplay (Minecraft threats, Doom combat)
• He explicitly cautions against assuming you can download a fly and watch it live an entire life:
  • Key behaviors (like reproduction) aren’t fully represented.
• The fly is not treated as fully sentient in this implementation—more like a puppet driven by mapped neural dynamics and incomplete sensory/motor mapping.

Featured gamers / sources (mentioned at the end of the video)

• Mutahar (channel creator)
• Google
• HHMI Janelia
• David (credited as the creator of “Fly Escape Room” / fly escape room project)
• GPT-6 Astra (mentioned in the loop used for turning models into a game)
• Blendy Remade
• Quad
• OpenWorm (project)
• Jacob Cokson (mentioned in relation to a viral thread about AI risk; also references the company making Claude)