Urgent.News

What's breaking now, across thousands of outlets.

AI

What’s Actually Inside 24,723 Tokens of a Search Result? We Broke It Down, Field by Field

Sponsored Content It's no secret that AI agents burn massive amounts of tokens on search results and file retrievals. They pull in...

What’s Actually Inside 24,723 Tokens of a Search Result? We Broke It Down, Field by Field

We haven't written up this one. Machine Learning Mastery has the full story — the link below goes straight to it.

Read the original at machinelearningmastery.com →

More in AI

The Navier–Stokes Problem Has Survived for a Century. Did AI Find Something New?

Imagine dropping a small stone into a perfectly calm lake. You see the ripples moving outward. It looks simple. Now imagine trying to predict the exact position of every ripple, every tiny swirl…

  • Navier–Stokes equations describe fluid motion factors like velocity, pressure, viscosity.
  • Century-old question: smooth fluid always remains smooth or solution breaks down.
  • AI claimed to solve problem in September 2026, sparking verification debates.

I Asked AI to Review My Code. It Found Bugs I Didn't See.

For a long time, I used AI in software development for one primary purpose: Write the code. Give it a requirement. Get some Python. Ask for a function. Get a function. Describe an API. Get an API.

  • AI reviewed author's code as senior software engineer
  • Identified potential bugs, edge cases, security concerns
  • Influenced author's coding process to anticipate problems

Agent Oversight Needs Metrics, Not Just Logs

Anthropic has published a snapshot of how it measures three parts of frontier AI development: the share of research led by AI, the oversight of internal agents, and the allocation of research compute.

Your AI Agent Has Too Many Permissions: A Practical Guide to Not Getting Burned

Somewhere right now , an AI agent has a standing API key with more access than the intern you'd never hand your prod credentials to.

  • AI agents often have excessive permissions beyond typical intern credentials.
  • Traditional security models don't account for the decision-making nature of AI agents.
  • Implementing short-lived, scoped tokens and human oversight can mitigate risks.

More from Friday 18 September →