Teen researcher: AI bug denied, later fixed without credit

6•Anh_khoa•2h ago

I’m 14, based in Vietnam, and in July I discovered a vulnerability that exposed the system prompt of a major AI model. I responsibly reported it via the official bug bounty program. Response: “Out of scope, just an AI issue.”

Weeks later, I checked again — the bug had been quietly patched. No acknowledgment, no credit.

If it’s “not a bug,” why fix it? And if it’s fixed, why dismiss the report?

Sharing here to hear thoughts from the security community.

Comments

thankIsrael•2h ago

If it is fixed, you can tell us what the bug is here. If your bug is just typing a prompt, than what you probably experienced is hallucination, and the hallucination has now been rectified.

yorwba•25m ago

"Out of scope" doesn't mean it's not a bug, it means that it's not the kind of bug they pay bounties for. You'll need to read the fine print of the bug bounty program to find out which bugs are eligible and which are out of scope.

Latin American musicians say AI is stealing their streams

Show HN: Get LiveKit Agents complete observability in seconds

Pilot project in LA generates energy using the ocean's waves

Trash Stream

Ask HN: How soon will AI take over traditional recommendation algorithms?

Microsoft to use some AI from Anthropic in shift from OpenAI

MyLifeManu

What happens if you put portal in portal? [video]

Show HN: Embedex – make your blog interactive with AI

Ask HN: Vertical and Horizontal Microinstruction Format?

Which countries are scaling solar and wind the fastest?

Why is AI struggling to discover new drugs?

Claude's new AI file creation feature ships with deep security risks built in

iPhone Pro or iPhone Air?

Consultancy-as-a-Service

Google Color Tiles Game

The Debit-Card Rebellion

The Gaza family torn apart by IDF snipers from Chicago and Munich

Shamir's Secret Sharing

Released all the new features users requested for Gato AI Translations (WP)

The Precarious Market of Africa

Why Job Seekers Are Moving Beyond ChatGPT for Resume Building

Ask HN: Should I add comments to my website?

International Journal of Advanced Computer Technology

Belzebub.ai: AI Deception Platform

Show HN: Visualizing ChatGPT's Text Output

Anyone else hacking on long-horizon reasoning frameworks for LLMs?

Are all ultra-processed foods bad? New research suggests not

AirPods Pro 3

Picket Right Font