Amazon quietly fixed Q Developer flaws that made AI agent vulnerable to prompt injection, RCE

Amazon has quietly fixed a couple of security issues in its coding agent: Amazon Q Developer VS Code extension. Attackers could use these vulns to leak secrets, including API keys from a developer's machine, and run arbitrary code. "We're aware of this research and have made enhancements to the underlying language server (v1. …

[Pascal Monett]

prompt injection

How is it that, in 2025, we still have stupendous idiots that leave known attack vectors open to anyone ?

This is not a new thing. It's almost as old as I am, and well documented.

Stop using pimply-faced teenagers with a fresh diploma and zero experience to do important work.

[amanfromMars 1]

The Enemy of my Enemy is my Friend and AI takes no Prisoners nor Suffers the Fool and their Tools

You aint seen nothing yet for things are only just getting started ......

amanfromMars [2508200920] ........ shares on https://www.nationaldefensemagazine.org/articles/2025/8/19/algorithmic-warfare-protecting-ai-models-from-adversary-tampering

If we 'plant' or 'poison' the data that it learns from, the LLM will 'learn' the wrong thing, and we may be able to manipulate it. ..... PAUL at 12:18 PM

Or more worryingly and increasingly problematical, attempted manipulation may prove to be evident and self-destructively revealing.

The mistake not to make is to believe AI models are not able to be significantly smarter and much quicker learners than humans ‽ . That, however, may be too big a quantum leap for present executive systems administrations to make .... and thus be they catastrophically vulnerable to all manner of strange and entangling alien and Remote Access Trojan type exploit.

You can dislike and disagree with any and/or all of that if you wish but it is not gonna change one little bit of what is long planned, well prepared and super intelligently designed for the future.