4 min read
EDITED 08 November, 2025: Updated with new developments
A shocking discovery has revealed that Foodpanda Pakistan may have been exposing sensitive information about its restaurant partners through an unprotected public API.
The issue surfaced when Amin Ahmed Khan, a local AI solutions architect, began building an experimental AI tool to analyze Foodpanda’s restaurant listings. This also included pricing, cuisines, and delivery times, etc.
However, instead of discovering patterns, he uncovered a serious data privacy lapse. On his LinkedIn, Khan wrote:
I was actually planning to build a small AI agent for fun, something that could analyze restaurants on Foodpanda Pakistan: pricing, ratings, cuisines, delivery times, and maybe even generate insights or recommendations.
So, I started exploring the public API to fetch some data for testing.
And what I found was alarming!
The unprotected endpoint, identified as pandora/vendors?country=pk, reportedly requires no authentication and has no rate limiting. The data it returns includes:
- Restaurant location coordinates
- Cuisine categories
- Delivery fees
- Owner contact details including phone numbers
- Vendor performance metrics
The exposure of this magnitude violates basic data protection principles and raises serious concerns over how Foodpanda handles sensitive vendor information. If the restaurant data is so casually accessible, it poses a serious threat to the users of the Foodpanda app, many of whom use their stored card information to buy items.
There is no need for an expert to verify that if this leak remains active, any nefarious element can potentially use the dataset to target restaurant owners directly, bypassing the need for traditional acquisition efforts.
“If any… new delivery platform enters Pakistan, they wouldn’t need to start from scratch,” the AWS expert warned. “They could build precise marketing strategies just by calling this API,” Khan warns.
The discovery reinforces a recurring theme in the tech industry: companies are racing toward AI-driven growth while overlooking fundamental data hygiene.
“Security isn’t about firewalls or encryption alone,” he added. “It starts with intentional data design. Every variable you expose matters.”
Although this is not a data hack in a traditional sense, it highlights serious negligence in API design and governance for a potential disaster. Leaving personal data unprotected erodes vendor trust and undermines the credibility of platforms handling user and partner information, especially for a platform as big as Foodpanda in Pakistan.
Khan has compiled and published the dataset for transparency on Kaggle. Some users commented that publishing the data like this was not the safest way to address this concern, to which Khan assured that all data set is masked for privacy.
In an interesting development, TechJuice did an independent verification of the facts, and learned that this breach had previously been identified by alandhar-based cyber security researcher Palvinder Singh. As later identified by Khan, Singh had approached Pandora (the parent company of Foodpanda) about the potential data leak, but the company disregarded his concern back then.
To Singh, Foodpanda replied the following:
At Foodpanda, we adhere to high standards of privacy and security. There has been no breach of data on the platform. As one of the country’s largest food-tech platforms, we are constantly striving to strengthen the experience for users, and our bug bounty program, the first of its kind in India’s technology ecosystem, is a step ahead in this direction. Through this, we are able to encourage users and technology enthusiasts to share constructive feedback and duly reward such enthusiasm. Such initiatives play a crucial role in establishing highest standards of data privacy and overall security of platforms.
Upon Khan’s notice, Pandora’s CTO approached him with the following statement:
We are currently investigating the report and would appreciate if you were able to remove the post and the dataset while we look into this matter. Some of the data you expose is required to be available while other might not.
The LinkedIn post has not been taken down as of writing this article, and neither is the Kraggle data deleted.
