Workplace apps are watching, keeping tabs, and sharing what they learn

The typical white-collar workplace in 2026 blends the personal and professional in ways previously unheard of. From BYOD (Bring Your Own Device) policies to the multitude of mobile apps required by many employers, personal data (including behavioral and location data) is increasingly finding its way into workplace systems. Even if only employer-provided devices are used for work, apps used to facilitate synchronous and asynchronous communication, as well as planning and organization, continue to have access to individuals’ personal data.

Collectively, these apps account for over 12.5 billion downloads on Google Play alone. Given that employees often have little choice but to install these apps for work, understanding their data practices is critical—users may be unknowingly exposing sensitive personal information, including contact details, financial data, and precise location, to their employer’s software stack.

On average, workplace apps collect around 19 data points and share approximately 2 data types. The three Google and Microsoft apps (Gmail, Google Meet, and Microsoft Teams) cluster at the top of the collection spectrum, each gathering 21–26 data types. Notion stands out as the app that shares the most data with third parties, sharing nearly half of what it collects. This is an aggressive pattern of data dissemination, especially given that the shared data includes personal identifiers and is used for advertising.

Key insights:

• Gmail is the most data-hungry app, collecting 26 distinct data types—more than any other app analyzed. It also collects data, including approximate location, app interactions, and user IDs, explicitly for advertising or marketing purposes.
• Notion shares the most data with third parties, distributing 8 distinct data types to third parties—including email addresses, names, user IDs, device or other IDs, and app interactions, several of which are shared for advertising or marketing. Sharing this data increases the likelihood of it being exposed in case of a breach.
• Six apps collect data explicitly for advertising or marketing: Gmail, Slack, Notion, Outlook, Todoist, and Zoom Workplace. The data types involved include personal identifiers like names, email addresses, user IDs, device IDs, approximate locations, and app interactions.
• Email addresses are collected for marketing or advertising by 3 out of the 10 investigated apps: Slack, Todoist, and Notion. 
• Zoom Workplace and Microsoft Teams collect precise location data, one of the most sensitive data types, and are the only two apps in this set to do so.
• Workday is the only investigated app that does not allow users to request deletion of their data, a significant gap for an HR platform that collects user IDs and location data.
• Zoom and Slack have experienced data breaches and Notion had vulnerabilities revealed that could have resulted in workers’ data leaks. Microsoft and Google, parent companies of several investigated apps, both also experienced data breaches in other products. Workday’s customer data was exposed by a third-party service provider. Only Todoist remains unconnected to any known data breaches.

Data collection and sharing practices and associated risks

To understand the risks of sharing that data with the developers or even third parties, Incogni’s researchers highlighted the following observations for select data types:

Data type	Potential risk in case of breach	Potential consequences when shared with third parties
Email Address	Exposed email addresses become primary targets for phishing attacks, spam campaigns, and credential-stuffing attempts.	Allows the shared address to be flooded with targeted marketing. Also, the email can be sold across data broker networks, making it nearly impossible to regain privacy.
Name	An exposed name, especially combined with other data points, can enable identity theft, social engineering attacks, and impersonation.	Allows personalized marketing that can feel invasive. Further, it helps advertisers build and enrich identity profiles that persist across platforms.
Location data	Exposed exact real-time or historical whereabouts can enable stalking, physical threats, and blackmail.	Allows for location-based profiling, meaning that advertisers can know exactly which stores users visit, how long they stay, and their commute patterns.
App Interactions	In-app actions can reveal work habits, communication style, and behavioral profiles.	Allows advertisers to understand a person’s preferences, activity patterns, and engagement levels to micro-target them with ads.
Device or Other IDs	Device IDs can be used to track and target specific devices for technological attacks and, when combined with other data, for social engineering attacks.	Allows advertisers to build a comprehensive profile of all activity on a given device. Persisting across every app that shares the same ID with the same ad network.

Gmail

Gmail is the most data-hungry app in the analysis, collecting 26 distinct data types, more than any other tool examined. It looks at approximate location, app interactions, and user IDs explicitly for advertising and marketing purposes.

In January 2026, a security researcher discovered a publicly accessible, unencrypted 96-gigabyte database containing approximately 48 million sets of Gmail account credentials.Google denied any internal breach, attributing the exposure to infostealer malware that harvests credentials from user devices. Meanwhile, a separate privacy controversy emerged when reports surfaced in late 2025 that Google had enabled its Gemini AI by default for Gmail users, allowing it to analyze private communications—often without explicit user consent.A class-action lawsuit in California followed, alleging Google intentionally obscured the opt-out process. For employees required to use Gmail for work, these incidents underscore how the app’s aggressive data collection practices create real and ongoing exposure.

Microsoft Teams

Microsoft Teams collects 25 data types, one of the highest counts in Incogni’s analysis. It’s one of only two apps in this set to collect precise location data, one of the most sensitive data categories available.

Microsoft Teams has also become a growing target for attackers. In 2025, Check Point documented how Microsoft Teams vulnerabilities enabled impersonation, spoofed notifications, and forged caller identities. Recognizing this shift, Microsoft announced that, beginning January 12, 2026, it would automatically enable a set of critical messaging safety protections for organizations still using standard configurations, designed to shield users from AI-driven phishing and malware attacks. While the security update is a step forward, it doesn’t change the volume of personal data the app collects, including precise location, which remains a concern for employees with little say over whether the app is installed on their devices. The platform’s deep integration into Microsoft 365 means data collected through Teams can feed into a much broader ecosystem of Microsoft services.

Zoom Workplace

Zoom Workplace collects 23 data types and, alongside Microsoft Teams, is one of only two apps in this study to collect precise location data—a category with serious privacy implications if exposed or misused.

Zoom also collects data for advertising, including personal identifiers such as names, email addresses, and user IDs. The platform has a documented history of privacy controversies: an earlier backlash over Zoom’s AI terms of service prompted significant policy clarifications, and Zoom updated its privacy statement in February 2026 to expand definitions of customer content and clarify recording access rights. More recently, a critical vulnerability, tracked as CVE-2026-22844, was discovered in Zoom Node Multimedia Routers, carrying a near-maximum severity score of 9.9, which could allow a meeting participant to execute remote code via network access. For employees who rely on Zoom for video meetings, the combination of precise location collection, advertising-linked data use, and a recurring history of vulnerabilities makes it one of the more complex apps to evaluate from a privacy standpoint.

Microsoft Outlook

Microsoft Outlook collects 22 data types as part of Microsoft’s broader 365 ecosystem, placing it among the most data-hungry apps in our analysis alongside its sibling product, Microsoft Teams. Like Gmail, Outlook collects data explicitly for advertising purposes, specifically device or other identifiers.

Outlook’s deep integration with the rest of Microsoft 365, including Teams, SharePoint, and Copilot AI, means data collected via email activity can inform a wide range of downstream processing. Microsoft has been rolling out a new version of Outlook to enterprise users, with full migration for Windows business plan users to be completed by April 2026, bringing expanded AI features that will process email content more extensively than before. For employees who have to use Outlook as their primary email client, these AI integrations increase the surface area of data analysis, even if Microsoft maintains that enterprise data is not used to train its models. Parent company Microsoft has experienced data breaches across other products, a reminder that the scale of its data holdings makes any security gap consequential.

Google Meet

Google Meet collects 21 data types and sits in the upper tier of our analysis alongside Gmail and Microsoft Teams.

As a Google product, Meet falls under the same parent company that, in mid-2025, suffered a major cyberattack that compromised a Google database managed through Salesforce’s cloud platform, exposing the contact details and business information of users across Google’s ecosystem. In late 2025, Google also enabled Gemini AI by default for Meet users, allowing it to analyze meeting content without requiring explicit consent from participants. While Google distinguishes between AI analysis for features like meeting summaries and that used to train its public models, the opt-out process has drawn criticism for being buried across multiple settings. For employees who join Meet calls without realizing they’re being processed by AI tools, the distinction between “helpful features” and “data collection” becomes meaningfully blurred.

Slack

Slack was shown to collect 17 data types (two of which are also shared). Email addresses, for example, are collected explicitly for advertising purposes—making Slack one of only three apps in Incogni’s study to do so.

The platform has had a troubled security track record: a previous breach exposed internal code repositories and, more recently, Japanese media conglomerate Nikkei disclosed in November 2025 that hackers used malware-stolen Slack credentials to access accounts belonging to over 17,000 employees and business partners, exposing names, email addresses, and internal chat histories.

Security experts have pointed to Slack’s lack of end-to-end encryption as a concern—workspace administrators and, under certain legal conditions, Slack itself retains technical access to message content. Workspace owners and administrators can access virtually all Slack communications, including messages employees consider private, such as direct messages between coworkers and conversations in private channels.For organizations that use Slack as a hub for sensitive internal discussion, the combination of advertising-linked data use and limited encryption makes it worth scrutinizing before treating it as a private communication channel.

Todoist

Todoist collects 17 data types, below the study average, but notably shares email addresses with third parties for advertising and marketing purposes, placing it among a small group of apps that use one of the most sensitive contact identifiers for commercial ends. Unlike several other apps in this analysis, Todoist has no known history of data breaches, making it the only app in this set with a clean breach record. However, the advertising use of email addresses means that simply having a work Todoist account could route your contact information into third-party ad networks. Todoist is made by Doist, a fully remote, independent company—a key structural difference from the large tech platforms in this study, though one that does not change the advertising data practices disclosed in its Play Store listing.

For employees using Todoist to manage work tasks, the app’s relative data minimalism is a plus, but the advertising-linked email sharing is a trade-off worth noting.

Trello

Trello collects 17 data types, among the least in the study. It’s owned by Atlassian, which also operates products like Jira and Confluence. The app does not appear to collect data explicitly for advertising purposes, distinguishing it from roughly half the apps in this study. 

In January 2024, data scraped from Trello surfaced for sale on a well-known hacking forum, totaling over 15 million records, including email addresses, names, and usernames. The data was obtained from a publicly accessible resource using email addresses from previous data leaks. Trello maintained that no unauthorized access had taken place.

Notion

Editor’s note: This section was updated to clarify distinctions between AI model training and third-party AI processing, and to provide additional context regarding AI-related data handling and disclosed security vulnerabilities.

Notion is a standout data-sharing app in our analysis, distributing 8 distinct data types to third parties, including email addresses, names, user IDs, and app interactions, several of which are shared for advertising purposes. This aggressive third-party sharing pattern increases the likelihood that a security vulnerability or incident at any one of those partners could lead to an unintended data leak. Furthermore, Notion’s privacy policy allows select third-party advertising technology partners to place tracking tools on users’ browsers to collect behavioral data—a practice that sits uncomfortably alongside Notion’s positioning as a secure, professional productivity workspace.

In late 2025, researchers showed how the LLM used by Notion was susceptible to data exfiltration using prompt injections. Through this method researchers were able to demonstrate that data unintended for the public was accessible to malicious actors. Notion did respond to the vulnerability disclosure and patched the exploit. In this case, there are no reported instances of the exploit being abused before it was patched. But this highlights the dangers associated with using emerging technologies in environments rich with personal or otherwise sensitive data. 

Importantly, for employees storing highly sensitive internal documents—like product roadmaps, HR notes, or client data—routing this information through any third-party AI provider inherently expands a company’s data exposure footprint.

Workday

Workday is the only app in our analysis that does not allow users to request deletion of their data, a notable data-privacy gap for an HR platform that sits at the center of employment records, payroll, and personal information. 

That policy takes on added weight given the company’s recent security history. In August 2025, Workday confirmed two related incidents tied to its use of Salesforce as a CRM platform, with attackers gaining access to business contact information, including names, email addresses, and phone numbers. The breach was part of a broader social engineering campaign linked to the hacker group ShinyHunters, which targeted multiple large enterprises using the same tactic. Workday stated there was no indication that core customer tenant data was accessed, but the incident raised broader questions about how enterprise HR platforms manage third-party integrations, and how they share private data with those parties.

For employees whose entire employment history (like performance data, compensation, location, etc.) lives within Workday, the absence of a data deletion right, combined with a supply-chain breach in the same year, makes this one of the more consequential apps to examine in any workplace privacy audit.

Methodology

The application sample was curated through a systematic review of industry-leading software evaluations and workplace technology benchmarks. By cross-referencing recurring recommendations and discussions about workplace requirements across prominent publications, the research team identified software frequently mandated or used in modern work environments in the US.

On March 20th, 2026, the research team collected information about their selected apps from the Google Play Store, noting the user information the developers claimed to collect and share, as well as the purposes they stated. The team also searched for reports of data breaches involving the applications, their developers, or connected entities, using several search engines and queries. 

For detailed information used in this study, see our public dataset. 

Use of visuals

Notwithstanding the terms of the CC BY-NC-SA 3.0 licenses of the visuals above, Incogni grants news organizations and other media entities permission to use the specified asset(s) in their news coverage or commentary, including on pages that display advertising.

The visuals can be downloaded or embedded using the menu at the top right of each visual. Embedded visuals preserve their interactivity.