If you’ve seen the series ‘Upload,’ then you know exactly where this is headed. Eon Systems, a San Francisco startup, just demoed the first complete adult fruit fly brain emulation connected to a physics-simulated virtual body. The digital insect walks, grooms, and forages entirely on its own. No reinforcement learning, no massive datasets. Just biological architecture, replicated one-to-one.

That's more than double the neuron count of any previous attempt. The model maps nearly 140,000 neurons and 50 million synaptic connections, predicting motor behavior with 95% accuracy. The AI simply mimics the real architecture of the brain.

The digital fly runs on a MuJoCo physics engine, meaning it experiences gravity and physical resistance. Its movements are indistinguishable from the real thing. Previous attempts focused on larval flies with far fewer neurons - this adult model is a massive leap.

Eon's next target is the mouse brain (roughly 70 million neurons), before eventually attempting human-scale emulation. Critics are already debating the implications - some see a path toward digital drug testing, others worry about digital consciousness. For now, it's proof that a biological brain can be replicated one-to-one in a machine.

The gap between a fruit fly and a human is enormous. But six months ago, nobody had done this with an adult insect either.

New York Senate Bill S7263 aims to ban AI chatbots from giving substantive advice in licensed professions like medicine, law, and engineering. It passed the Senate Internet and Technology Committee unanimously and is heading to a full floor vote.

The bill targets the companies that deploy chatbots, not the developers who build the models. If passed, it would impose civil liability on anyone who lets their AI provide guidance that normally requires a professional license. Users who suffer damages could sue directly.

Critics call it protectionism designed to shield high-paid professionals from competition. Others say "substantive" is too vague, leaving AI companies guessing about what they can safely provide. If it passes, it could set a precedent for other states, potentially forcing companies like OpenAI and Google to geofence features for New York residents.

The uncomfortable question is whether this is really about consumer protection or about keeping AI out of industries that charge $500 an hour.

🤖 Minimax, Moonshot, and Zhipu race to launch autonomous AI agents

🔭 ESA researchers find 800 undocumented cosmic anomalies using AI

🛡️ Anthropic sues government over supply chain risk designation

🎼 German court to hear game-changing copyright case against Suno AI

🛍️ OpenAI cancels plans for direct shopping inside ChatGPT

🧪 Andrej Karpathy releases open source framework for autonomous research

Here’s how to try it:

🖱️ Step 1: Pick the right model Open ChatGPT and choose Thinking from the model picker. In ChatGPT, Thinking = GPT-5.4 Thinking on paid tiers that have manual model selection.

🧠 Step 2: Start agent mode Open the tools menu and select agent mode, or type /agent in the chat box. That’s the official way to give ChatGPT a task that involves browsing, research, and actions across websites.

🖥️ Step 3: Give it a real multi-step task Ask it to do something concrete, like: “Find three flight options for a weekend trip and put them in a comparison table.” Agent mode can navigate sites, gather information, and organize the result while pausing if it needs confirmation or clarification.

Then see how it does. Did it finish cleanly, or did it wander into the weeds?

Is the emulation of a fruit fly the first step toward digital immortality, or just a very complex science experiment? And, should AI be legally barred from giving professional advice? See you next time! 🚀

Stay informed, stay curious, and stay ahead with Jumble!

Zoe from Jumble