The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Back to feedPost
This work demonstrates how AI outputs can appear valid while still being unsafe to act on under real conditions.
It focuses on the gap between:
– what a system produces
– and what a system is actually allowed to execute
The analysis highlights:
– how outputs can pass evaluation but fail at execution
– why risk emerges at the transition from intent to action
– where systems lack constraints at the point of commit
The goal is to identify where AI-driven decisions become actionable without being fully validated against current state, authority and conditions.
The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Related posts
How good of a web designer is AI?
I put Claude Design to the test asking it to redesign my client's product inventory page I built in Framer.
In under 10min it produced 2 variations. Neither can go on client's website directly but it surfaced two years' worth of overlooked improvements I can implement manually.
The gap between AI output and production-ready website is there, but in the hands of a web designer who knows what to do with the output, it has a valuable place in the workflow.
Feels like AI isn’t replacing the designer, it’s just speeding up the "thinking layer". Like you said, it surfaces things we somehow miss for months.
Still needs a human eye to make it actually usable, but as a brainstorming partner, it’s kind of crazy good.
Built an AI Security Scanner for GitHub Orgs in 2 Weeks
Engineered an end-to-end AI platform that scans entire GitHub organizations for vulnerabilities and auto-generates fix PRs.
What it does:
Crawls 100+ repos via GitHub GraphQL API in batch mode
Runs GPT-4 over real source code to detect SQL injection, XSS, hardcoded secrets, CSRF, auth bypasses
Scores by severity, generates secure code fixes, creates PRs automatically
Real-time WebSocket dashboard shows live progress
Tech: Next.js 15 · Node.js · Supabase · BullMQ · Redis · OpenAI · CrewAI
Open source, MIT licensed: github.com/aymandakir/org-code-ai
Shipped in under 2 weeks. Full stack, production-ready.
Great work! And it's open source too. 😄
A short presentation of my entry for Impossible UI with Rive entry of SPEN project.
And a bonus content, the watcher! An AI being, watching all!
Thank you all for your likes and comments so far, from the start to the end of the project!
🔗 Full project: https://www.behance.net/gallery/247442001/SPEN-Silent-Predictive-Enforcement-Network-FUI-HUD
🔗 Contra Submission P1: https://lnkd.in/dxhJ5ckU
🔗 Contra Submission P2: https://lnkd.in/db8qKD4T
🔗 Contra Submission P3: https://lnkd.in/dChAer6W
🔗 Contra Submission P4: https://lnkd.in/d54h7_aq
🔗 Contra Submission P5: https://on.contra.com/zpmB6r
I love this, can you pls explain how to do this just briefly
Trending
Claude
Claude has entered the design space. How are you using Claude Design?
Contra University
Learn from expert creatives how to earn more using next-gen AI tools.
creativeaiflow
Creative AI workflows are evolving. What tools do you use, and what are their strengths and weaknesses?
portfolioreview
The best portfolios tell a story, not just show a grid. Share yours for feedback.
freelancerlife
Freelancer life is wins, pivots, and everything in between. What’s yours right now?