Someone at BrowserStack is leaking users' email addresses
TL;DR Highlight
A developer using unique emails per service discovered that an email used only with BrowserStack was passed to a third party via Apollo.io, and BrowserStack has not responded.
Who Should Read
Developers operating SaaS services and integrating external CRM or marketing tools, or developers interested in privacy. Specifically, teams using BrowserStack or adopting sales automation platforms like Apollo.io.
Core Mechanics
- The author uses a method of generating unique email addresses for each service, allowing them to track which service leaked their email. They received spam to a unique email used only with BrowserStack.
- When asking the spammer for the source, they were told it came from Apollo.io. After contacting Apollo.io, they initially lied, claiming the email was derived from a combination of publicly available information and corporate email patterns (firstname.lastname@domain).
- However, the email address was in a completely unique format that could not be inferred algorithmically. After being challenged by the author, Apollo.io eventually admitted that BrowserStack shared the contact information through its 'customer contributor network' on February 25, 2026.
- Apollo.io's 'customer contributor network' operates by companies using Apollo sharing their customer contact data on the Apollo platform. This feature is enabled by default and must be explicitly opted-out to disable.
- The author attempted to contact BrowserStack multiple times but received no response. They ironically noted that the email contained the phrase 'No spam, we promise!'
- Three possible causes for the leak were identified: BrowserStack regularly sells or provides user data, a third-party service used by BrowserStack collects and leaks the data, or BrowserStack internal employees or contractors steal the data.
- From a GDPR perspective, even if Apollo claims GDPR compliance, BrowserStack would need a valid legal basis to share customer emails from its support database with Apollo, which it likely lacks. Both BrowserStack and Apollo are potentially in violation.
Evidence
- A commenter familiar with Apollo.io's operation pointed out that 'this isn't a data hack, it's how Apollo originally works.' They explained that if Apollo customers don't opt-out, their customer contact data is automatically shared on the Apollo network. (See https://knowledge.apollo.io/hc/en-us/articles/20727684184589)
- There was also an opinion that the BrowserStack sales team likely used Apollo for CRM purposes without realizing the privacy implications and uploaded the entire customer list. It's common for general sales representatives to proceed without understanding the consequences of such data sharing.
- A comment from a GDPR expert analyzed that Apollo's claim of 'Legitimate Interests' as a legal basis is not applicable to sharing emails from BrowserStack's support database with Apollo. They pointed out that both BrowserStack and Apollo should report GDPR violations and take preventative measures.
- A comment stated that the '+tag' style email alias (e.g., name+browserstack@gmail.com) is no longer effective because many services already perform de-aliasing to extract the actual email. Creating a completely separate inbox with a unique email is more effective for tracking.
- A comment mentioned that BrightData (headless browser service) recently had a similar data leak incident. The commenter discovered that URLs accessed only through their headless browser fingerprinting project were later crawled by Anthropic's ClaudeBot, suggesting that attackers may be using Claude to analyze stolen customer data.
- Shared experiences indicated that Seamless.io operates in a similar manner. A commenter discovered a very personal email address in the Seamless.io system and suspected a colleague of leaking the address book, but didn't know how to prevent it.
- A case study from UK's Compare The Market shared a similar situation. Two unique emails using different domains started receiving spam simultaneously on the same day, but no action was taken after reporting it, citing 'cannot be proven.'
How to Apply
- If you are using BrowserStack and also using sales platforms like Apollo.io, immediately check and disable the 'customer contributor network' data sharing opt-out in Apollo settings. Data sharing is enabled by default, so you must explicitly opt-out to prevent customer contacts from being automatically shared on the Apollo network.
- Teams planning to adopt external CRM or sales automation tools (Apollo, Seamless.io, etc.) must review the platform's data sharing policies and opt-out structure before contracting. Specifically, check the terms and conditions regarding how customer contacts uploaded to the platform are shared on the platform network.
- If you want to track leaks on a personal level using unique emails per service, using completely separate email aliases (e.g., SimpleLogin, Fastmail aliases) is more effective than the '+tag' method (name+site@gmail.com). Many services automatically remove '+tags' to extract the actual email.
- If you operate a service subject to EU GDPR, sharing email addresses from your customer support database with external marketing/sales platforms is not justified by 'Legitimate Interests' and may violate GDPR. You must clarify the processing basis with your legal team before sharing data.
Terminology
Related Papers
Show HN: Mindwalk – Replay coding-agent sessions on a 3D map of your codebase
Claude Code나 Codex 같은 AI 코딩 에이전트가 세션 중 코드베이스의 어떤 파일을 탐색하고 수정했는지를 3D 지도 형태로 시각화해서 재생해주는 로컬 도구다. 에이전트가 작업을 어떻게 이해했는지 한눈에 파악할 수 있다.
Ghost Font: A font that humans can read but AI cannot
움직임(모션)을 이용해 글자를 표현해서 AI 모델이 정적 이미지 분석으로는 메시지를 해독하지 못하게 막는 실험적 프로젝트인데, 커뮤니티에서는 이미 GPT-5.6, Claude Opus 등으로 해독에 성공한 사례가 속출해 실효성 논쟁이 뜨겁다.
GPT-5.6, Grok 4.5, Claude, and Muse Spark build the same 4 apps
12개 LLM 모델에게 레이캐스터 미로, 루빅스 큐브, 계산기, Game of Life 앱을 각각 5번씩 만들게 해서 성공률·비용·속도를 비교한 실전 벤치마크다. GPT-5.6 Sol이 전반적으로 가장 일관된 결과를 냈고, Grok 4.5는 가성비 면에서 눈에 띄었다.
Benchmarking coding agents on Databricks' multi-million line codebase
Databricks가 자사 실제 코드베이스를 기반으로 여러 AI 코딩 에이전트의 성능과 비용을 직접 측정했고, 모델 토큰 가격과 실제 태스크 비용이 전혀 다르다는 점, 그리고 오픈소스 모델이 이제 최상위 수준에 도달했다는 점을 확인했다.
Estimating Uncertainty from Reasoning: A Large-Scale Study of Multi- and Crosslingual MCQA Performance in LLMs
LLM이 저자원 언어 질문을 받을 때 영어로 추론하게 하면 불확실성 추정 성능이 고자원 언어 수준으로 올라온다.
LLM-as-a-Verifier: A General-Purpose Verification Framework
LLM의 토큰 확률 분포를 활용해 discrete 점수 대신 continuous 점수를 뽑아내면, 추가 학습 없이 코딩·로봇·의료 에이전트 평가 정확도를 SOTA로 끌어올릴 수 있다.