logicallee

I am curious about the state of prompt injection attacks on frontier models. Are they still vulnerable? For example, is it safe to let Claude Code look at user-submitted data if it also helps manage some of the infrastructure or code? Can they just be asked to identify prompt injection attacks and flag and ignore them, or do injection attacks change the models' behavior despite the owner's prompts? What are best practices?

Here on Hacker News some time ago (like years ago) I saw a link to criticism of something that became an Internet standard, the format of the link was either text or a picture of text. It was a very long time ago, like 60s or 70s. It was a formal, and negative, review. Among the criticisms was that it wasn't considered that novel.

As I remember it, this was at Bell Labs and the protocol later became TCP/IP, but I might be filling in the blanks in my memory incorrectly.

What I'm sure of is that it was critical, at a research lab of some kind, included the statement that the proposal wasn't novel (others did similar things etc), and I recall that the author went on to create an Internet standard out of it later, such as TCP/IP or something similarly important.

Thanks for helping me track down this link.