Four cybersecurity companies were breached through a login that was not theirs.
In June, four companies that sell cybersecurity for a living had customer records taken out of their CRM. None of them were broken into.
The intruder got into a competitive intelligence vendor whose app connects into Salesforce. From there they took the OAuth tokens that connection runs on and used them to appear as the vendor inside every customer tenant on the other side. To the CRM it looked like an approved integration doing its job.
What the vendor confirmed
They detected the intrusion the same day and published their own findings rather than waiting to be reported on. The first step was a compromised legacy credential. Their word, in their own disclosure.
Their response list:
- Credentials revoked.
- Unauthorized code removed.
- External forensics engaged.
- Law enforcement and customers notified.
Salesforce disabled the integration five days after detection. An extortion group claimed the data and set a deadline.
Read the entry point again. Legacy. Something old enough that the word made it into the public account, still live, still holding reach.
What it cost
The exposed records were subscription details, products trialed, business contacts and marketing history. For a security vendor that is not a customer list, it is the commercial map of the business, and it is worth most to the people selling against it.
No affected company has published a loss and I am not going to invent one. The number exists and each of them can compute it: the value of the renewals and open opportunities described in those records, multiplied by the share that now get negotiated against someone holding the pricing and the trial history.
The part that generalizes
Those four companies could have counted every account in their directory, rotated every password and enforced MFA on every employee. The outcome would be identical, the identity that opened the door was never in their directory, it belonged to a vendor, it was granted once during procurement, and after that it persisted without anyone revisiting it.
Which brings me to what I said I would cover this week.
Rotating that legacy credential would have made this less likely. Deleting it would have made it impossible.
Hardening acts on likelihood. Rotate a secret, enforce MFA, narrow a network path, add monitoring. Each one lowers the chance an identity is taken and none of them changes what that identity reaches once it is taken. The worst case stays exactly where it was.
Deletion acts on reach. Revoke a grant, retire an account, remove a trust relationship. The worst case gets permanently smaller, whether or not anyone ever tries.
Both are worth doing. They are not interchangeable and they do not cost the same. A board shown a single remediation list sorted by severity rating cannot tell which items bought a smaller worst case and which only bought a less likely one.
Three questions worth putting to your board this quarter
- How many third party integrations hold standing read access to our customer records, and who approved each one?
- If one of those vendors is compromised tomorrow, what is the largest thing an attacker reaches through a token we issued ourselves?
- On our current remediation plan, which items reduce the chance of compromise and which reduce what is reachable, and what is the dollar difference between them?
If the third question has no answer, the first two are interesting rather than useful. A board cannot govern an exposure it cannot price.
Next week: what a dormant account is actually worth. The cheapest exposure to remove is almost always the one nobody is using, and almost nobody prices it before deleting it.
The Identity Risk Play-by-Play. One real identity incident a week: how a login opened the door, what it reached, and what the exposure was worth. Also published at artemion.io/newsletter.
← Back to the issue catalogue