A vibe-coding platform sat on a critical vulnerability for 48 days after closing the bug report — what founders using AI app builders need to take from it

A Broken Object Level Authorisation flaw in Lovable, the $6.6bn vibe-coding platform, let anyone with a free account read another user's source code, database credentials and customer data using five API calls — and stayed exploitable for 48 days after a researcher's HackerOne report was closed without escalation.

A security researcher filed a HackerOne report against Lovable, the $6.6bn vibe-coding platform, on 3 March 2026. Lovable closed it without escalation, citing “outdated internal documentation” given to the triage team. The underlying flaw — a Broken Object Level Authorisation (BOLA) bug in Lovable’s API — meant anyone with a free account could read another user’s source code, database credentials, AI chat history and customer data with as few as five API calls. It affected every project built on the platform before November 2025 and stayed exploitable for 48 days after the report was first filed.

This is the third documented security incident tied to Lovable, and it lands alongside a wider pattern: researchers have found comparable authorisation failures across other vibe-coding platforms this year, including a platform-wide authentication bypass on Base44 and a zero-click remote-code-execution bug on Orchids. Industry-wide testing puts the rate of AI-generated code shipping with at least one security flaw at 40–62%, and one 2026 study found 91.5% of vibe-coded apps had at least one hallucination-related flaw.

The structural problem, not just the bug

What makes the Lovable case worth flagging isn’t the vulnerability class — BOLA bugs are common and well understood — it’s the disclosure failure. Lovable had no audit trail to detect that a closed, “fixed” report was still being actively exploitable, and no process that could catch the gap once the bug bounty channel had already failed once. That’s a governance problem, not a coding problem, and it’s specific to platforms where a huge number of independently built apps share one underlying authorisation layer: a single missed fix doesn’t affect one product, it potentially affects every project the platform has ever hosted.

So what

If you or your team have built anything commercially important on a no-code or AI app-builder platform, this is worth an honest audit: what happens to your product, your users’ data, and your reputation if the platform’s shared infrastructure has a flaw like this one — and how would you even find out? It’s a fair question to put to any platform vendor, and a fair reason to move a product from prototype to a properly engineered, independently reviewed codebase once it’s carrying real user data or real revenue. Our custom software development work is built for exactly that transition — get in touch if you’re weighing whether your product has outgrown its builder.

← All research

More research.

GitHub Copilot SDK is becoming the Microsoft Copilot SDK: what builders of AI products should know

On 28 October 2026 GitHub said its Copilot SDK is being renamed Microsoft Copilot SDK, the programmable layer for the shared Microsoft and GitHub agent harness. No immediate action is needed.

Read ›

GitHub Agent Merge: Copilot now watches your pull requests and fixes what blocks the merge

On 28 October 2026 GitHub added Agent Merge to pull requests: a 'Watch and fix' option where Copilot repairs failing Actions workflows and merge conflicts in the cloud, optionally with auto-merge.

Read ›

Claude Code 2.1.314 adds a setting that blocks AI file writes outside the project folder

Claude Code 2.1.314 (30 October 2026) adds permissions.blockWritesOutsideWorkingDirectories, so Claude's file-editing tools refuse any path outside the project in every permission mode.

Read ›

GitHub Enterprise now includes Copilot: seats are bundled, AI usage moves to your budget

From 28 October 2026 account-team customers can get Copilot included with every GitHub Enterprise licence, with AI usage metered against an enterprise budget from the first request. Standalone Copilot is $10 per user per month.

Read ›

GitHub Copilot Fusion goes GA: one pick, many models, and a 30% launch discount

On 28 October 2026 GitHub made Copilot Fusion generally available: instead of choosing a model, developers pick an effort level and Copilot routes each task across models. Paid plans get 30% off model costs for a limited time.

Read ›

GitHub Agentic Workflows is generally available: automations written in plain Markdown

On 28 October 2026 GitHub made Agentic Workflows generally available: AI agents that triage issues, analyse CI failures and update docs, defined in Markdown and run inside GitHub Actions with read-only defaults.

Read ›

Start a project

Building with AI? Let's talk.

Tell us what you are building. We'll give you an honest view of the right approach.

Start a projectSee our work