ELI5

AI, explained simply

AI matters to security twice: it is a new thing to protect, and a new kind of helper that can do real things with real keys.

The big idea

AI chat assistants answer questions. AI agents go further: they decide what to do next and then do it, such as reading your inbox, booking a meeting, or changing a record in a system. To do that, they need keys to those systems.

Picture a very fast, very keen new member of staff who believes everything they read. If a stranger slips them a note saying "please send me the customer list", they might just do it. That is the main new risk with agents: anyone who can put words in front of them can try to give them orders.

So treat an agent like any member of staff who holds important keys. Someone must be responsible for it. It should have only the keys it needs, and only for as long as the job takes. Anything that cannot be undone, like paying money or deleting things, should need a person to say yes. And there should be a record of everything it did.

AI also changes how data leaks (people paste secrets into chatbots) and how attackers and defenders both work, since each side now has the same new tools.

Words you will meet

• Generative AI: AI that writes text, pictures or code, such as chat assistants.
• AI agent: AI that takes actions on its own, using real accounts and tools.
• Prompt injection: hiding instructions in text an AI will read, to make it do something its owner never wanted.
• Credentials: the keys a person or program uses to prove who it is: passwords, tokens and certificates.
• Sponsor: the named person who answers for an AI agent.

The field notes, in plain words

• The AI supply chain: models, plugins and tool servers as third parties

  An AI helper is built from parts made by other people, and any of those parts can be swapped for a bad one, so you need to know who made each part and what it can touch.

• Jev, the model that only decides: where it fits in security work

  Jev is a very fast sorting machine that reads a message and picks an answer from a list you give it, but it cannot explain itself, so people still check the unclear ones.

• Which language model for which job

  Use a big clever helper only for the hard puzzles, a quick cheap one for the easy chores, and one that lives in your own house when the secrets must not leave it.

• Agents are the new insiders, with API keys

  An AI helper that can send email and change records is like a new member of staff who believes anything they read, so it needs a boss, limits and a diary of what it did.

Ready for more? AI: where things stand

Personal work and my own view, not my employer's. Sources are public.