| Document Title |
|---|
| Pinterest fired engineers who tracked layoffs in Slack — what it says about privacy, trust, and internal telemetry | |
|
|---|
| Pinterest fired engineers after scripts tracked layoffs via internal tools; why employment-status data is sensitive and how to prevent internal surveillance. | |
| Title Attribute |
|---|
| oEmbed (JSON) | |
| oEmbed (XML) | |
| JSON | |
| View all posts by Admin | |
| How Apple’s Lockdown Mode can derail iPhone forensics — and why that’s the point | |
| Senators grill Waymo and Tesla on robotaxi safety — what’s actually at stake | |
| Page Content |
|---|
| Pinterest fired engineers who tracked layoffs in Slack — what it says about privacy, trust, and internal telemetry | |
| Nature | |
| Climate | |
| / | |
| General | |
| / By | |
| Admin | |
| Pinterest reportedly fired two engineers after they wrote scripts to identify which coworkers were being removed from internal tools during a layoff — and then shared that list more broadly. On the surface, this is a workplace drama story. Underneath, it’s an unusually clear case study in how modern companies actually run: identity systems as the source of truth, chat platforms as de facto org charts, and “internal data” that is technically accessible long before it is socially acceptable. | |
| The incident matters beyond Pinterest because the same ingredients exist in almost every tech company: centralized identity, Slack or Teams, HR systems, and a long tail of internal dashboards and APIs. When times are calm, nobody thinks too hard about the line between observability and surveillance. When layoffs happen, that line lights up. | |
| In this explainer, we’ll look at what likely happened, why it’s tempting to do this kind of tracking, where it crosses ethical and policy boundaries, and what organizations can do to reduce both privacy harm and the urge for shadow information-gathering. | |
| What happened (and what “a script” probably means here) | |
| According to the BBC, Pinterest said “two engineers wrote custom scripts improperly accessing confidential company information to identify the locations and names of all dismissed employees and then shared it more broadly,” calling it a policy violation and a privacy issue for affected staff. The reporting also describes the mechanism as watching for employee names being removed or deactivated inside an internal communication tool “like Slack.” | |
| In many companies, Slack (or similar) is tied directly to the identity provider (Okta, Azure AD, Google Workspace, etc.). When an account is disabled, a cascade follows: access tokens expire, groups change, and the user stops appearing in certain directory searches, channels, and integrations. If you have API access (even read-only), you can often infer who was terminated simply by detecting state changes: | |
| A user disappears from the “active users” list. | |
| A user’s profile becomes deactivated. | |
| A bot can no longer DM them. | |
| Their membership changes across channels or user groups. | |
| A “script” in this context doesn’t have to be sophisticated. It could be a few dozen lines of code polling an API, comparing yesterday’s user list to today’s, and emitting an alert. From a purely technical perspective, it’s the same pattern engineers use for legitimate operational tasks: diffing two snapshots to detect change. | |
| The difference is what is being detected (people), why it’s being detected (layoffs), and where the results go (shared broadly). | |
| Why employees do this during layoffs | |
| There’s an uncomfortable truth about layoffs: people usually learn the shape of the event through side channels before leadership clarifies anything. Sometimes that’s because leadership can’t share details yet. Sometimes it’s because “we’re still working through the details” is a euphemism for “we don’t want to say.” | |
| So employees reach for whatever signals exist: | |
| Friends suddenly go silent. | |
| Calendar invites vanish. | |
| Access to repos is revoked. | |
| Slack status flips, or the person disappears from the directory. | |
| Tracking those signals can feel like self-defense. People want to know: | |
| Is my team impacted? | |
| Did my manager get cut? | |
| Are my closest collaborators still here? | |
| Is the company okay, or is this a larger restructuring? | |
| That motivation is human and predictable. But predictable behavior can still be harmful behavior. | |
| The privacy problem: layoff status is sensitive information | |
| A termination event is not just “work trivia.” It’s sensitive personal information about someone’s employment status, often tied to benefits, immigration, health insurance, and future job prospects. | |
| Even if a company plans to announce a headcount reduction publicly, the identity of the individuals and their locations is typically meant to be disclosed on a need-to-know basis: | |
| HR and payroll need details. | |
| IT needs to execute offboarding. | |
| Legal needs to ensure compliance. | |
| Managers need to communicate directly with their teams. | |
| Broad internal sharing of a list of terminated employees is different. It can: | |
| Remove the affected person’s ability to control the narrative. | |
| Expose someone’s location or team affiliation. | |
| Encourage gossip and speculation (“was it performance?” “was it political?”). | |
| Increase the risk of targeted harassment or doxxing outside the company. | |
| Pinterest’s framing — that it violated former colleagues’ privacy — is not just PR. It’s a real category of harm. | |
| The security problem: access control isn’t the same as authorization | |
| Many internal systems work on coarse permissions: if you’re an engineer, you might be able to query a directory or use an internal API. That doesn’t mean you’re authorized to use it for every purpose. | |
| This is where a lot of organizations struggle. They build internal tools that are: | |
| Easy to use, | |
| Powerful, | |
| Poorly governed. | |
| And then they rely on policy (“don’t do that”) as the primary guardrail. When the pressure is high, policy-only guardrails fail. | |
| NIST SP 800-53 is one of the standard catalogs organizations use to think about control families like access control and auditing. Even without getting lost in control IDs, the basic idea applies cleanly here: data access should be constrained, monitored, and attributable to legitimate business purposes — especially for sensitive categories of information. | |
| In other words: “you can technically read this” should never be treated as “it’s fine for you to read this.” | |
| The cultural problem: Slack has become the org chart | |
| Most companies now have two parallel realities: | |
| The formal reality: HR systems, reporting lines, official announcements. | |
| The lived reality: Slack channels, group DMs, GitHub mentions, on-call rotations. | |
| When something changes in the formal system (like offboarding), it immediately produces visible artifacts in the lived system. Employees interpret those artifacts as truth — sometimes more strongly than they trust leadership communications. | |
| That mismatch creates a perverse incentive: | |
| If leadership won’t tell you what’s happening, | |
| you will reconstruct it from whatever telemetry leaks. | |
| This incident is a reminder that “internal transparency” is not just a comms strategy — it’s also an information-security strategy. If people feel they must piece together reality from leaks, they will. | |
| Where the engineers crossed the line | |
| Even if you empathize with why someone might build such a script, there are at least three bright lines that get crossed: | |
| 1) Purpose limitation | |
| If the data source is “confidential company information,” the expectation is that it’s used for a legitimate business function, not for layoff reconnaissance. | |
| NIST’s Privacy Framework emphasizes managing privacy risk and using practices that protect individuals. A practical translation is “collect and use data for specific, legitimate purposes, and avoid secondary uses that create new harms.” | |
| A script to identify terminated colleagues is almost definitionally a secondary use: the offboarding signals exist to protect systems and execute HR processes, not to generate an internal layoff list. | |
| 2) Amplification | |
| People notice disappearances in Slack organically — that’s ambient information leakage. | |
| A script turns ambient leakage into a structured dataset (names, locations, likely teams, time of termination). That is amplification: the harm potential rises sharply when vague signals become a clean list. | |
| 3) Redistribution | |
| Sharing the output “more broadly” is the step that makes it hard to defend as mere curiosity. It creates a new distribution channel for sensitive information and makes the authors accountable for downstream misuse. | |
| What companies can do: reduce leakage, increase trust, and tighten controls | |
| There’s a misconception that the solution is “lock everything down.” In practice you need three complementary moves: governance, technical controls, and communication. | |
| 1) Treat offboarding events as sensitive and design for privacy | |
| Offboarding inevitably changes systems, but you can reduce the informational exhaust: | |
| Minimize public-facing directory changes until communications occur. | |
| Avoid mass user removals that are easy to diff. | |
| Consider delaying certain non-security-critical updates by hours so they don’t act as a real-time layoff feed. | |
| The goal isn’t to hide reality forever. It’s to avoid turning a painful event into a scavenger hunt. | |
| 2) Add purpose-based access controls and logging | |
| If internal APIs can reveal employee status changes at scale, then: | |
| Access should be scoped to roles that need it. | |
| Bulk export should require justification. | |
| Queries should be logged with identity and intent. | |
| Automated polling should stand out. | |
| This is where the “audit and accountability” mindset matters: if a script is enumerating users and emitting alerts, it should trigger detection. | |
| 3) Have a humane comms plan for layoffs | |
| The biggest driver of shadow tracking is uncertainty. | |
| Companies can reduce the impulse to scrape internal tools by being explicit: | |
| When will impacted employees be told? | |
| When will teams be informed? | |
| What can be shared, and when? | |
| Where should people go for verified updates? | |
| If leadership provides timely, specific information, the “need” for DIY lists drops. | |
| 4) Give employees a sanctioned way to check on collaborators | |
| This is subtle but important. People are not only curious — they’re trying to coordinate work and check on friends. | |
| A simple, sanctioned directory status message (“this account is no longer active”) without timestamps, location, or lists could satisfy basic needs without enabling mass reconstruction. | |
| A wider trend: layoffs as an information-security stress test | |
| Layoffs reveal weak points in governance because they create: | |
| a burst of sensitive events, | |
| a high emotional temperature, | |
| and a lot of access churn. | |
| That’s exactly when you see edge cases: employees scraping systems, managers improvising, and tools being used in ways nobody designed for. | |
| Sites like Layoffs.fyi exist because people want an independent signal about the scale of cuts in the industry. Inside a company, that same need exists — except the signals are more direct and the stakes are personal. | |
| Bottom line | |
| Pinterest firing engineers for scripting layoff tracking isn’t just “don’t be nosy.” It’s a warning that internal observability can become internal surveillance the moment organizational trust drops. | |
| If your tooling makes it easy to turn identity churn into a list of terminated coworkers, people will do it — especially during layoffs. The fix isn’t only punishing the people who wrote the script; it’s building systems and communication practices that don’t turn offboarding into a data leak, and that treat employment status as the sensitive information it is. | |
| Sources | |
| https://www.bbc.com/news/articles/cn0k670n0ydo | |
| https://layoffs.fyi/ | |
| https://www.nist.gov/privacy-framework | |
| https://csrc.nist.gov/pubs/sp/800/53/r5/final | |
| ← | |
| Previous Post | |
| Next Post | |
| → | |
| oEmbed (JSON) | |
| oEmbed (XML) | |
| JSON | |
| View all posts by Admin | |
| How Apple’s Lockdown Mode can derail iPhone forensics — and why that’s the point | |
| Senators grill Waymo and Tesla on robotaxi safety — what’s actually at stake | |
| Pinterest fired engineers after scripts tracked layoffs via internal tools; why employment-status data is sensitive and how to prevent internal surveillance. | |
| |
日本語
English
العربية
Čeština
Dansk
Nederlands
Eesti
Suomi
Français
Deutsch
Ελληνικά
Magyar
Italiano
한국어
Latviešu valoda
Lietuvių kalba
Norsk bokmål
Polski
Português
Română
Русский
Slovenčina
Slovenščina
Español
Svenska
Türkçe