Hacker News

“But it appears 1 or more organizations have successfully jail-broken Fable 5”

This is hardly true or it’s true of all frontier models and this was only magnified by Fables capabilities. It’s that you could hand Fable 5 vulnerable code, ask it to fix it, return patch plus test cases proving the fix and exploit relevant detail falls out as a byproduct of legitimate secure code review work.

I challenge anyone to provide a fix for this “exploit” without compromising Fable’s ability to patch unsecure code.

I run orchestrated agents(context warm up, fork-join, file based communication, set of trailofbits skills, some dba skill, some sourcegraphindex, and few mine, API skill). None of skills are specific to out codebase(even my), some are explicitly for Claude.

Finds 90% bugs using Codex and Gemini 200usd subs.

You do not need service for auditing or review.

Just overlay you ci with ai.

I do not even sast (because we use rust typestates and invariants; and effect ts; and failfast).

Have seen death of 5 services for audit and review killed by "just overlay ci with ai in max sub on latest model".

As with so much (LLM) security work, the devil is in the details: "~25 security issues per codebase" means nothing without a grounding in the codebase's actual security model, capabilities exposed to an attacker, etc. I haven't used Aikido's product, but my experience with similar tools is that tend to not find actual security issues until a proper security model is introduced for grounding.

(I say this as someone who is, broadly, extremely impressed by and interested in the use of LLMs for security research.)

I hope it’s an improvement on their current PR review code scanning, which alerts on code that only looks possibly vulnerable in isolation, without looking at the context. I guess I assumed it was an LLM being extremely lazy, but maybe it’s just static analysis. Anyway it’s pretty annoying.

This looks promising, but I find it a little odd to bury the bulk of plan limitations under "fair-usage limits". When the limitations are specifically coupled to plans, it feels less like an FUP and more like plan-specific caps that should be surfaced more directly.

We’ve been using aikido code scanning and pen test tools and been pretty impressed. Will have to take a look at this.

This is marketed as a defensive tool, but how do you prove that you check against "your" source code?

Looks like a solid bridge between SAST and manual review. Will check it out.

[flagged]