In late July 2026, attackers reached internet-facing programmable logic controllers at more than 30 Minnesota water and wastewater utilities, changed their passwords and IP addresses, and locked operators out of monitoring and control (Tenable). The FBI and EPA reported that water systems in seven states were hit (Western Water). The weakness behind these attacks is the same one exposed in February 2021, when the BBC reported that someone had remotely changed the chemical settings at a water treatment plant in Oldsmar, Florida.
Oldsmar is still the most cited case study for water sector cyber risk, and it is still disputed. Its lessons apply directly to the 2026 attacks: control systems reachable from the internet, protected by credentials that can be shared, guessed or changed, with no reliable way to verify who is issuing commands.
The federal recommendations issued after the July attacks are the same basic controls that would have applied at Oldsmar: disconnect control devices from the public internet, use secure gateways for remote access, use strong unique passwords and maintain the ability to operate manually.
On February 5, 2021, a plant operator in Oldsmar watched his cursor move across the screen as someone raised the sodium hydroxide setting from about 100 parts per million to 11,100 parts per million. He reversed the change. Pinellas County Sheriff Bob Gualtieri later told reporters that “someone hacked into the system, not just once but twice” (NPR). A Massachusetts advisory for public water suppliers documented the conditions at the plant (Industrial Cyber):
In April 2023, former Oldsmar city manager Al Braithwaite called the incident a “non-event” and said it was likely caused by a plant employee. The FBI told CyberScoop that it “was not able to confirm that this incident was initiated by a targeted cyber intrusion of Oldsmar.” The Pinellas County Sheriff’s Office said the case remained open.
The unresolved outcome points to the real weakness. When every workstation uses the same remote access password and sits directly on the internet, there is no reliable way to separate an outside session from an inside one. Investigators could not give a definitive answer to the most basic question: who changed the setting?
|
|
Oldsmar, 2021 |
U.S. water utilities, 2026 |
|
Entry point |
TeamViewer on SCADA computers connected directly to the internet |
PLCs connected directly to the internet |
|
Credential weakness |
One remote access password shared across all computers |
Device passwords that attackers could change to lock operators out |
|
Effect |
Sodium hydroxide setpoint changed; reversed by an operator |
Loss of monitoring and control; some utilities moved to manual operations |
|
Attribution |
FBI could not confirm an outside intrusion |
Linked to Iranian-affiliated actors; Minnesota link described as preliminary |
EPA data collected between the two events shows the same conditions were widespread:
PKI was never a workable option for most OT environments. Certificate authorities need network connectivity. Certificates need renewal on schedules that small utility staffs cannot sustain. A PKI handshake takes 300 to 700 milliseconds, and many PLCs and RTUs cannot run a certificate stack at all.
Without a practical way to give each device and session its own cryptographic identity, utilities used what was available: shared logins, vendor remote desktop tools and flat networks. Those choices were made for operational reasons. They also leave the operator unable to prove whether a command came from an authorized session.
AKMSecure’s Autonomous Key Management™ delivers encryption and authentication where PKI was never viable. Each condition documented at Oldsmar and in the 2026 attacks maps to a specific AKM capability:
|
Condition documented in 2021 and 2026 |
How AKM addresses it |
|
Shared, default or changeable passwords protecting control access |
Every session is independently verified with keys that refresh per session. There is no standing credential to share, reuse or steal. |
|
Workstations and PLCs connected directly to the internet |
AKM is air-gapped capable. Authentication does not depend on a certificate authority or an outside service. |
|
No reliable way to separate an outside session from an inside one |
Every packet is independently verified. Traffic from an unverified source is rejected at the protocol layer. |
|
Legacy, resource-constrained control equipment |
A sub-1MB embedded executable and a handshake under 1ms fit controllers that cannot run a certificate stack. |
|
Equipment lifecycles measured in decades |
The symmetric-key architecture is quantum-resilient, so protection does not expire with the next cryptographic transition. |
AKM does not stop an authorized operator from entering a wrong value. Setpoint limits and process alarms cover that risk, and Oldsmar had pH alarms in place. What AKM provides is Zero Trust at the protocol layer: no implicit trust in any connection and no persistent credentials. In the 2026 attacks, reaching a PLC over the internet and changing its password was enough to lock operators out. Under AKM, commands must arrive through an independently verified session, so network access and a password are not enough to issue them. AKM also supports the zone-and-conduit segmentation that IEC 62443 describes.
AKMSecure delivers a patented Autonomous Key Management™ protocol built to replace outdated PKI approaches with a dynamic, quantum-secure, air-gapped-capable architecture. Instead of relying on persistent credentials that can be stolen, reused, or abused, AKM enables independently verified sessions with no standing privileges left behind. The result is a model that better aligns with Zero Trust principles, reduces certificate-based risk, and supports resilient operations across enterprise IT, OT and Tactical Edge environments. Built to NSA-grade security standards and deployable as a lightweight SDK, AKMSecure helps organizations modernize trust at the protocol layer without rebuilding everything around it.