Lovable

Lovable is a Swedish AI‑coding startup that provides an AI‑assisted development platform. The platform lets users write code, interact with an AI chatbot, and store their projects in a cloud‑based environment. Users can choose to make projects public or keep them private, and the service offers a free tier for individuals. Collaboration features enable teammates to view and edit shared code while receiving real‑time suggestions from the AI.

The company’s headquarters are located in Sweden, as stated in the incident description. Being described as a startup indicates an early‑stage organization with a limited operational footprint. No further details about employee count, revenue, or geographic reach are provided in the source material.

Lovable’s distinguishing attribute is its tight integration of generative AI assistance directly within a code‑editing workflow. This specialization targets developers who seek instant AI‑driven code completions, debugging help, and natural‑language explanations alongside their source files. The platform’s design emphasizes project visibility controls, allowing users to toggle between public showcases and private workspaces. Security experts have noted that the firm’s approach highlights the importance of secure defaults and thorough threat modeling for AI‑assisted tools.

On 19 April 2026 a security issue was reported when a user discovered that free accounts could view other users’ code, AI chat histories, and associated customer data. Lovable initially claimed that the visibility of public projects was an intentional feature, but after criticism admitted the setting was a mistake. The company subsequently restored private chat functionality for projects that had been marked as public, acknowledging the error. While some users appreciated the later transparency, others characterized the initial response as gaslighting, and security commentators framed the episode as a case of missing secure defaults rather than a traditional breach.

The incident underscores the challenges AI‑coding startups face in balancing openness with data protection. It also illustrates how user‑facing settings can unintentionally expose sensitive information when default configurations are not rigorously vetted. Lovable’s experience serves as a concrete example for the broader industry of why threat modeling must extend to collaborative AI features in development environments.