Proofpoint says it found evidence of compromise at 64% of the European organizations it watched being hit by targeted attacks on cloud accounts. Researchers released the numbers on Thursday. The figure raises an awkward question for any provider that manages email and identity for business customers: who actually has the power to stop an intruder once a login has been stolen?
Start with a typical ticket. A provider deletes a malicious email and closes the case. The customer, meanwhile, may still have an active incident, because the attacker has already signed in. At that point the evidence lives in sign-in records, not in the mailbox, and those records may belong to a different supplier or team. A company paid to filter email might lack both the permissions and the contractual right to end a session inside the customer’s tenant.
The scale is notable, although the numbers need careful reading. Proofpoint counted 4.5 times as many phishing detections in Europe as a year earlier, against 1.5 times globally. Those are detections, and the 64% figure covers outcomes only among organizations already singled out for targeting. Neither figure shows whether attacks rose, monitoring improved or the customer mix changed. A budget for incident response needs a different measure: how many cases required someone to act.
Responsibility splits awkwardly in a managed Microsoft 365 setup. One supplier may run email security, another may watch identity alerts, and the customer may keep approval over anything that interrupts access. At noon, that costs a few phone calls. At 2 a.m., an analyst spots the alert, opens a ticket and waits for someone with authority to wake up.
Giving the provider standing power to act carries a cost as well. Revoking a session can lock out a legitimate user, and a hasty move on a shared or privileged account can disrupt far more than one mailbox. Buyers should decide in advance which actions a provider may take alone and which need a signoff. They should also check where a contract’s response clock stops, at notification or at containment.
Connected apps add another layer. Proofpoint found more than 188,000 AI-enabled applications linked through authentication systems. That counts connected apps, not malicious ones. Still, restoring email access doesn’t answer what permissions a user already granted, and if the contract ends at mailbox recovery, that review lands back on the customer.

