Four tech stories in plain words: an AI-run mouse surgery, falling AI prices, drone data tools, and AI that hunts software bugs.
This week, an AI agent ran parts of a surgery on a mouse, and AI prices kept dropping fast. A drone company also launched new inspection software, and several big tech firms now offer AI tools that hunt for security bugs. You do not need to do anything today. Each story is worth watching, because it shows where everyday technology may go next.
An AI agent and a mouse surgery
An AI agent is a program that can make choices and take steps on its own, not just answer questions. A biotech startup called Exin Therapeutics says it used one to do brain surgery on a mouse, as reported by Tom's Hardware.
Here is what the company describes:
- The mouse was a model of epilepsy, a condition that causes seizures. The goal was to reduce seizures.
- The agent picked one of 19 gene therapies from a spreadsheet. It also chose the brain area, using three research papers.
- It drove robotic arms that injected 250 nanoliters, a very tiny amount, into the brain.
- A licensed vet approved every step that could not be undone. The vet also held a kill switch and opened the skull, a step the AI was not allowed to do.
The team used Anthropic's Sonnet model. The report says a newer model, Opus 5.5, refused the task once it saw it was a biological experiment. The mouse was reported to be recovering well. Tom's Hardware notes the work raises ethical questions. These are the company's own claims, shared on X.
What to watch: whether outside experts review the work, and what rules form around AI in medicine.
AI is getting much cheaper
AI companies charge by the "token," which is a small piece of text, like part of a word. Prices for tokens are falling fast. Tom's Hardware, citing The Register, reports that Anthropic's price per million tokens fell 77% in under five months. It went from $2.65 in May to $0.62 in October 2026.
This affects a network called Darkbloom. It lets people rent out their unused Mac computers to run AI work for others. The top 50 members earned a median of about $159.86 a month. After the price drop, that may fall to about $36.77 a month, or $1.23 a day. Darkbloom says it is looking for new kinds of work and new clients.
The same report notes that some companies still spend more on AI than planned. The reason is that they use far more tokens, not that each token costs more.
What to watch: cheaper AI may show up as more AI features in apps you already use.
Drones that check power lines
Many power lines, pipelines and plants are checked by drones and sensors. Percepto, a company that makes self-flying inspection drones, launched software called Axion, reports The Robot Report.
Axion is meant to work with data from many brands of drones and sensors, not just Percepto's. The company says it uses more than 160 AI models to spot problems. Examples include a fallen wire or a part that is running too hot. Workers can also ask questions in plain language. Percepto says it will keep selling its drones and will make them in the U.S.
Why it matters: the company says the goal is to catch equipment problems before they grow. For ordinary people, that is about the systems that keep lights on and fuel flowing.
AI tools that hunt security bugs
A security bug is a flaw in software that an attacker could use. According to MarkTechPost, Anthropic, OpenAI, Cloudflare and Google now each offer an AI agent that searches code for these flaws.
All four use a similar idea. One agent finds a possible bug. A separate agent tries to prove it is not real. Only the bugs that pass this check get a suggested fix. This matters because Google's security team writes that careless AI scanning can be right less than 7% of the time.
Two more points from the report:
- Anthropic says it built its tool partly because attackers were already using AI to find flaws.
- Some tools only report findings by default. People still review and approve the fixes.
What to watch: if these tools work well, the apps and websites you use could get safer updates faster.
Have a question about anything here? Email me at sales@appwt.com or call (888) 565-0171, and I will answer it.
Sources
- AI agent used to perform controversial brain surgery on a mouse, Anthropic Sonnet controlled motorized surgery rig, Tom's Hardware, 2026-10-31
- Firm that lets you rent out your idle Mac for AI inference workloads warns of plummeting token costs, Tom's Hardware, 2026-10-31
- Percepto launches Axion AI platform to unify inspection data analysis, The Robot Report, 2026-10-30
- AI Security Agents Compared: Claude, Codex, Mantis, Cloudflare, MarkTechPost, 2026-10-30
Tags
Frequently Asked Questions
Did an AI really do brain surgery on a mouse by itself?
Not fully. The company says the AI chose the treatment and brain area and drove the robotic arms. A licensed vet approved each step that could not be undone, held a kill switch, and opened the skull.
What is a token in AI?
A token is a small piece of text, like part of a word. AI companies charge based on how many tokens are used.
How much did AI prices drop?
A report cited by Tom's Hardware says Anthropic's price per million tokens fell 77%, from $2.65 in May to $0.62 in October 2026.
What does Percepto's Axion do?
It is software that reads inspection data from many brands of drones and sensors. The company says it uses AI to spot problems like a fallen wire or an overheating part.
Do AI bug hunters fix code on their own?
The tools can suggest fixes, but some only report findings by default. People review the results before changes are made.
Enjoyed this article?
Share it with your network
