๐ฌ๐งThe Register - AI/MLโขStalecollected in 6m
Lovable Denies Leak, Blames HackerOne

๐กAI coding startup's botched security response lesson for builders
โก 30-Second TL;DR
What Changed
Free accounts could access others' credentials, chat history, source code
Why It Matters
Undermines trust in Lovable's security for AI coding users. Highlights risks of poor multi-tenant isolation in SaaS AI tools. May deter developers from similar vibe-coding platforms.
What To Do Next
Audit your AI coding tool's tenant isolation to block cross-account data access.
Who should care:Developers & AI Engineers
Key Points
- โขFree accounts could access others' credentials, chat history, source code
- โขLovable attributed exposure to 'intentional behavior' and unclear docs
- โขCompany shifted blame to HackerOne bug-bounty service
- โขResponse criticized as mishandling vulnerability reports
๐ง Deep Insight
AI-generated analysis for this event.
๐ Enhanced Key Takeaways
- โขThe vulnerability stemmed from an Insecure Direct Object Reference (IDOR) flaw within Lovable's API endpoints, which failed to properly validate user authorization tokens when fetching project metadata.
- โขHackerOne issued a public statement clarifying that their platform's triage process was not bypassed, noting that Lovable's internal security team had independently closed the report as 'Informative' before the public disclosure occurred.
- โขSecurity researchers highlighted that Lovable's initial response violated standard Coordinated Vulnerability Disclosure (CVD) practices by publicly disparaging the reporter's methodology rather than addressing the underlying architectural flaw.
๐ Competitor Analysisโธ Show
| Feature | Lovable | Cursor | Bolt.new | Replit Agent |
|---|---|---|---|---|
| Primary Focus | Vibe-coding/No-code | AI Code Editor | Web-based AI App Dev | Full-stack IDE/Agent |
| Pricing Model | Freemium/Subscription | Subscription | Open Source/Cloud | Subscription |
| Security Posture | Under scrutiny (IDOR) | Enterprise-grade | Community-driven | Enterprise-grade |
๐ ๏ธ Technical Deep Dive
- โขThe vulnerability was identified as an IDOR (Insecure Direct Object Reference) vulnerability in the
/api/v1/projects/{project_id}endpoint. - โขThe API lacked server-side ownership verification, allowing any authenticated user with a valid JWT to access project objects by simply incrementing or guessing the UUID of other users' projects.
- โขThe platform utilizes a multi-tenant architecture where project data is stored in a shared PostgreSQL instance, and the application layer failed to enforce Row Level Security (RLS) policies for cross-tenant data access.
๐ฎ Future ImplicationsAI analysis grounded in cited sources
Lovable will face increased scrutiny from enterprise customers regarding their SOC2 compliance status.
The mishandling of a critical vulnerability disclosure typically triggers mandatory security audits and delays enterprise procurement cycles.
The company will be forced to implement a formal Bug Bounty program policy update.
Public backlash regarding the blame-shifting toward HackerOne necessitates a transparent revision of their vulnerability disclosure policy to regain developer trust.
โณ Timeline
2024-09
Lovable launches its AI-powered 'vibe-coding' platform to the public.
2026-04
Security researcher reports critical IDOR vulnerability to Lovable via HackerOne.
2026-04
Lovable publicly denies the vulnerability and shifts blame to the bug-bounty platform.
๐ฐ
Weekly AI Recap
Read this week's curated digest of top AI events โ
๐Related Updates
AI-curated news aggregator. All content rights belong to original publishers.
Original source: The Register - AI/ML โ