New research: The State of SaaS, October 2026. Read the report

Breach Radar · Live tracker

SaaS Data Breaches, Day by Day

Every confirmed incident at software we track, each with the vendor's own notice.

14
incidents
14
companies
6
via a third party
Nov 20, 2026
last updated

Incidents by Day

9 incidents in the last 30 days, 4 through an outside partner, app or AI tool.

Oct 24, 2026Nov 22, 2026
Company's own systemsOutside partner, app or AI tool

Confirmed incidents only, dated by the vendor's disclosure, searched back to Aug 22, 2026. Source: vendor notices and news reports, HissingKitty Breach Radar.

Latest Incidents

  1. Nov 20, 2026

  2. Nov 19, 2026

    • Optimizely

      an unauthorized third party exploited a flaw in the Optimizely Web Experimentation account signup flow. They used it to impersonate users on targeted customer accounts and insert malicious JavaScript into those customers’ projects. This was not a credentials breach. The incident was limited to Web Experimentation. Less than 5% of our customers were impacted. No other Optimizely product was affected.

      Source: Company notice
  3. Nov 17, 2026

    • Optimizely

      Malicious JavaScript added to some Web Experimentation customer projects and served to their site visitors on Nov 16; researchers counted 188 infected scripts at 178 organizations

      Source: News report
  4. Nov 13, 2026

    • Bitrise

      Unauthorized access to an internal system that Bitrise says does not store or process customer data; it found no evidence customer data, secrets or source code were exposed.

      Source: Company notice
    • Cursorvia Fireworks AI

      Coding agent traces (mostly from 2025, sessions not in Privacy Mode) copied from a bucket shared with Fireworks AI; less than 0.01% of customers had API credentials exposed. With those, the attacker reached archived non-production source code, limited support data and emails, usage metrics and contact details, some agent conversation summaries and some credentials users sent to support.

      Source: Company notice
  5. Nov 10, 2026

    • Ubervia Fireworks AI

      Names, addresses, dates of birth and driver's license numbers of close to 150,000 drivers, stored with Fireworks AI. Attackers used stolen Fireworks internal credentials to reach Uber's data on October 7.

      Source: Press
  6. Nov 8, 2026

  7. Nov 3, 2026

    • GitBook

      For about 3.5 hours on Nov 2 and 3, the attacker could reach customer data: API tokens, integration tokens and secrets, session signing secrets, questions visitors asked GitBook Assistant, documentation sites behind sign-in, connected GitHub and GitLab repositories, and customer content. GitBook treats all customer data as exposed. It found no evidence that other customers' content was modified. Billing details, passwords and IP addresses were not affected.

      Source: Company notice
  8. Oct 30, 2026

    • Northflank

      Customer environment variables and secrets, which Northflank told customers to assume were decrypted; users reported unauthorized use of AI API keys.

      Source: Company notice
  9. Oct 21, 2026

    • Railway

      A subset of customer environment variables, mainly AI provider keys, for at most 1.85% of workspaces (about 100,000), plus linked tokens for 1,852 users

      Source: Company notice
  10. Oct 19, 2026

    • Genesys

      Business and account information in its internal learning system and CRM, including names, business emails, phone numbers, job titles, company and account details. Genesys says Genesys Cloud, Engage and customer data in those platforms were not accessed.

      Source: vendor
  11. Oct 17, 2026

    • Microsoftvia ConnectWise Cloud Backup

      A stolen ConnectWise Cloud Backup app credential was used to sign in to customer Microsoft 365 tenants as the app, with no MFA prompt; Huntress counted at least 162 compromised tenants.

      Source: press
    • OpenCode

      Third-party AI provider keys (OpenAI, Anthropic, Google) saved in OpenCode Console workspaces, for 1.4% of Console users, after unauthorized access to a legacy system.

      Source: Press
  12. Aug 29, 2026

    • Intercomvia Canny (feedback app)

      An access key linking Canny to customers' Intercom support systems was stolen and used to read support data, including names, emails, phone numbers and some support messages.

      Source: press

Every 2026 Incident

DisclosedCompanyThird partyWhat was exposedSource
Nov 20, 2026ShopifyTriple Whale (analytics app)Store customer names, emails, phone numbers, mailing addresses and total spend

Triple Whale found unauthorized access to internal systems on Oct 31, 2026 and contained it Nov 1; it notified affected merchants. Sneak Energy said access to its store data ran Oct 30 to Nov 1, with no payment details, order details or passwords accessed, and that it was one of more than 65,000 companies affected. Triple Whale has not said how attackers got in.

Company notice
Nov 19, 2026OptimizelyNone namedan unauthorized third party exploited a flaw in the Optimizely Web Experimentation account signup flow. They used it to impersonate users on targeted customer accounts and insert malicious JavaScript into those customers’ projects. This was not a credentials breach. The incident was limited to Web Experimentation. Less than 5% of our customers were impacted. No other Optimizely product was affected.

We detected it, blocked delivery of the malicious code, and closed the vulnerability within three hours of detection. We also removed the injected code from all affected projects. No malicious code remains in our systems. We have contacted every Web Experimentation customer directly to confirm whether they were affected. If you do not use Web Experimentation, you do not need to do anything. If you

Company notice
Nov 17, 2026OptimizelyNone namedMalicious JavaScript added to some Web Experimentation customer projects and served to their site visitors on Nov 16; researchers counted 188 infected scripts at 178 organizations

The attacker used a flaw in the Experimentation account signup flow. Optimizely fixed it, removed the injected code, ended sessions, revoked tokens and notified affected customers.

News report
Nov 13, 2026BitriseNone namedUnauthorized access to an internal system that Bitrise says does not store or process customer data; it found no evidence customer data, secrets or source code were exposed.

Bitrise took all services offline on Nov 13 and restored them gradually through Nov 16. It advises customers to rotate all secrets, access tokens and certificates stored in Bitrise.

Company notice
Nov 13, 2026CursorFireworks AICoding agent traces (mostly from 2025, sessions not in Privacy Mode) copied from a bucket shared with Fireworks AI; less than 0.01% of customers had API credentials exposed. With those, the attacker reached archived non-production source code, limited support data and emails, usage metrics and contact details, some agent conversation summaries and some credentials users sent to support.

The access key Cursor issued to Fireworks was exposed by Fireworks. The same attack led to a malicious copy of the Claude Code extension in Cursor's marketplace on October 29. Cursor contacted each affected customer directly; rotate any credentials shared with Cursor support or pasted into agent sessions.

Company notice
Nov 10, 2026UberFireworks AINames, addresses, dates of birth and driver's license numbers of close to 150,000 drivers, stored with Fireworks AI. Attackers used stolen Fireworks internal credentials to reach Uber's data on October 7.

Fireworks told Uber on October 9; Uber cut its data connections with the vendor and filed notices with several U.S. attorneys general in November. Uber's alerts did not fire because the attackers used Fireworks' own credentials. Affected drivers are being notified by letter.

Press
Nov 8, 2026NotionFireworks AINotion AI chat transcripts used for its Insights feature, from fewer than 1% of workspaces

Chats carried no user or workspace labels and no tool call results. Notion turned off Insights and stopped using Fireworks.

Company notice
Nov 3, 2026GitBookNone namedFor about 3.5 hours on Nov 2 and 3, the attacker could reach customer data: API tokens, integration tokens and secrets, session signing secrets, questions visitors asked GitBook Assistant, documentation sites behind sign-in, connected GitHub and GitLab repositories, and customer content. GitBook treats all customer data as exposed. It found no evidence that other customers' content was modified. Billing details, passwords and IP addresses were not affected.

GitBook detected the incident on November 2, immediately contained it, took the app and authenticated docs offline, rotated every credential and secret, invalidated all customers' API tokens, fixed the SSO flaw, and restored services on November 3. They advised customers to rotate any existing GitBook API tokens and review connected credentials/integrations.

Company notice
Oct 30, 2026NorthflankNone namedCustomer environment variables and secrets, which Northflank told customers to assume were decrypted; users reported unauthorized use of AI API keys.

Northflank patched the exploit within an hour of a Wiz alert, worked with providers to revoke credentials and advises rotating all secrets and setting IP restrictions and budget limits. It found no evidence customer workloads were accessed.

Company notice
Oct 21, 2026RailwayNone namedA subset of customer environment variables, mainly AI provider keys, for at most 1.85% of workspaces (about 100,000), plus linked tokens for 1,852 users

Railway paused builds, rotated credentials, reset every session and token, and moved builds to single-tenant machines. Full report published Nov 10, 2026.

Company notice
Oct 19, 2026GenesysNone namedBusiness and account information in its internal learning system and CRM, including names, business emails, phone numbers, job titles, company and account details. Genesys says Genesys Cloud, Engage and customer data in those platforms were not accessed.

Genesys contained the intrusion by Oct 23 and suspended the learning system. Expect phishing that uses real account details.

vendor
Oct 17, 2026MicrosoftConnectWise Cloud BackupA stolen ConnectWise Cloud Backup app credential was used to sign in to customer Microsoft 365 tenants as the app, with no MFA prompt; Huntress counted at least 162 compromised tenants.

ConnectWise shut down app access. Microsoft 365 admins should disable the affected ConnectWise apps and review credentials on every app in the tenant.

press
Oct 17, 2026OpenCodeNone namedThird-party AI provider keys (OpenAI, Anthropic, Google) saved in OpenCode Console workspaces, for 1.4% of Console users, after unauthorized access to a legacy system.

OpenCode emailed affected users on Oct 17 to revoke and replace those keys with each provider. Keys kept in the local command-line and desktop apps were not involved.

Press
Aug 29, 2026IntercomCanny (feedback app)An access key linking Canny to customers' Intercom support systems was stolen and used to read support data, including names, emails, phone numbers and some support messages.

Canny cut the connection and invalidated the key on Aug 28. Intercom customers who connected Canny should rotate the key and review support access logs.

press

Read the full story: Notion AI Chats Exposed in Fireworks AI Breach. More in Security news.

What to Do When a Vendor Has an Incident

  • Read the vendor's notice and check whether your account was contacted.
  • Rotate any passwords, API keys or tokens that could have been in the exposed data.
  • Ask for the vendor's sub-processor list, so you know which third parties hold your data.
  • Put security terms on the agenda for your next software renewal.

Methodology and Limits

  • We list an incident only when the vendor or a regulator has confirmed it in writing, and every row links that source.
  • Only companies we track are included. The date is when the vendor confirmed the incident.
  • "What was exposed" uses the vendor's own words, shortened. We don't guess scope or blame.
  • A nightly check finds new notices; each one is reviewed before it is added.

More in our Methodology. Missing an incident or spot a mistake? Tell us.

Common Questions

What counts as a security incident here?

A breach or data exposure that the vendor itself or a regulator has confirmed in writing. Rumors, social posts and unconfirmed claims are left out.

Why list third-party vendors?

Many incidents now start at a supplier, such as an AI model host. Naming the third party shows where your data actually went.

Does an incident mean I should switch tools?

Not by itself. Look at how the vendor responded: how fast it disclosed, what it shut off and what it told customers to do.

How often is this updated?

A nightly check looks for new notices at the companies we track. Each find is reviewed before it is added.

Cite this tracker

HissingKitty, Software Security Incidents 2026, updated Nov 20, 2026, https://hissingkitty.com/breach-radar