Industrial infrastructures rely on real-time insights, unfettered data flows, and the seamless orchestration of diverse operational technologies. Few platforms are as pivotal in this ecosystem as AVEVA’s PI Web API, a powerful portal that bridges operational data with enterprise applications and dashboards. But with this criticality comes constant risk—a reality underscored by the recent disclosure of a cross-site scripting (XSS) vulnerability, tracked as CVE-2025-2745, shaking assumptions about the platform’s security under certain circumstances.
The newly reported vulnerability, affecting all PI Web API versions up through 2023 SP1, is rated at a CVSS v4 score of 4.5 and v3.1 score of 6.5. While these scores don't place it among the highest-severity vulnerabilities, the nuance of its exploitation vector, and where the risk manifests, deserve a deep-dive. According to the CISA advisory and AVEVA’s own disclosures, this is a classic case of improper neutralization of input during web page generation—commonly referred to as Cross-site Scripting (CWE-79).
What distinguishes this flaw from the litany of XSS issues in the past is the interplay between user privilege, browser content security policy (CSP) enforcement, and social engineering. Notably, successful exploitation requires:
In the case outlined by AVEVA and independent researchers, the attacker has to:
The notable risk factors in practice are:
For administrators and security professionals, the call to action is unmistakable:
Above all, the lesson is one of continuous learning and adaptation. Today’s mitigations are only as strong as tomorrow’s weakest link—and in the world of critical infrastructure, the cost of complacency is simply too high.
Source: CISA AVEVA PI Web API | CISA
A Closer Look at the AVEVA PI Web API Vulnerability
The newly reported vulnerability, affecting all PI Web API versions up through 2023 SP1, is rated at a CVSS v4 score of 4.5 and v3.1 score of 6.5. While these scores don't place it among the highest-severity vulnerabilities, the nuance of its exploitation vector, and where the risk manifests, deserve a deep-dive. According to the CISA advisory and AVEVA’s own disclosures, this is a classic case of improper neutralization of input during web page generation—commonly referred to as Cross-site Scripting (CWE-79).What distinguishes this flaw from the litany of XSS issues in the past is the interplay between user privilege, browser content security policy (CSP) enforcement, and social engineering. Notably, successful exploitation requires:
- Authenticated access with creation/update permissions for annotations or the ability to upload media files.
- Persuading a target user to disable their browser’s content security policy protections—a feat most likely accomplished via social engineering tactics.
- Rendering annotation attachments from within a browser, specifically when the security postures are relaxed.
Technical Analysis: Breaking Down the Exploit Chain
To appreciate both the risk and the mitigations, a brief technical breakdown is in order.Attack Surface
The PI Web API’s annotations feature allows users to upload attachments—documents, images, even SVG or PDF files—tying them to a historical data stream or event annotation. In scenarios where file extension allowlists are poorly configured, or if a user’s browser fails to enforce or is manually tricked into disabling CSP, an attacker can embed arbitrary JavaScript in a file. Upon another user opening that annotation attachment in the browser, the script executes with their session’s privileges.In the case outlined by AVEVA and independent researchers, the attacker has to:
- Gain authenticated access and privileges to create or update annotation objects (not uncommon in multi-user scenarios).
- Upload a specially crafted attachment containing executable JavaScript (e.g., within an SVG image).
- Convince a user (often through targeted phishing or manipulation) to disable the content security policy—say, by providing instructions during a support interaction, referencing performance, or offering a “fix.”
- The victim, often with elevated access of their own, opens the attachment, triggering the script.
Potential Impact
While the baseline risk is dampened by these prerequisites, the implications remain serious:- Executed JavaScript could siphon session tokens or inject further malicious payloads, leading to lateral movement within secure operational environments.
- Attackers could alter data, exfiltrate sensitive telemetry, or manipulate process visualization—risks particularly menacing in critical manufacturing and industrial contexts.
- For organizations that routinely share dashboards or annotation data across business and production domains, even a single successful exploit could open a persistent avenue for further compromise.
- Network-based access.
- High complexity (a barrier against widespread, automated exploitation).
- No required attack technology beyond privilege and user interaction.
- Localized scope, with moderate confidentiality and integrity impact.
Real-World Context: Where Are Organizations at Risk?
PI Web API is entrenched in numerous critical infrastructure environments worldwide—spanning power grids, oil refineries, chemical plants, and beyond. According to AVEVA’s disclosures, its adoption permeates sectors described as “critical manufacturing,” but in practice, deployments stretch far wider, mirrored by OSIsoft’s historic global reach prior to its acquisition by AVEVA.The notable risk factors in practice are:
- Legacy environments: Many PI systems are managed by OT (operational technology) teams, where patch management and rapid security response can lag that of IT systems.
- Shared access: Privilege “creep” can give more users than ideal the ability to create annotations or upload attachments, increasing the probability of an internal threat (intentional or otherwise).
- Flat network architectures: In less mature security programs, PI Web API endpoints may be reachable from both business and operations segments, compounding the fallout of lateral movement.
Response and Mitigations: What AVEVA and CISA Urge
Both AVEVA and cybersecurity authorities recommend a multi-pronged risk mitigation strategy, reflecting longstanding best-practices for industrial control systems (ICS):1. Patch Promptly
The most critical action: upgrade to PI Web API 2023 SP1 Patch 1 or newer. This patch addresses the XSS vulnerability directly, closing the door to known exploit vectors. Administrators can retrieve patched releases via the OSIsoft Customer Portal by searching for relevant version updates.2. Harden Attachment Handling
- File Extension Allowlist: AVEVA documentation urges a review and tightening of the file extensions permitted in annotation attachments. Removing high-risk types (notably SVG and PDF) significantly reduces the scope for payload delivery.
- Policy Enforcement: IT and security policies should forbid—both technically and operationally—the disabling of browser security features, particularly CSP, in production contexts. User education here is paramount, as social engineering remains a key step in successful exploitation.
- RESTful Access: Rather than rendering annotation attachments inline in browsers, organizations should access these files via direct REST API requests, bypassing potentially unsafe browser rendering altogether.
3. Principle of Least Privilege
Careful audit and revocation of unnecessary privileges is essential. Only trusted, properly vetted users should have “Annotate” access rights, reducing the number of accounts from which attacks could originate.4. Network Segmentation and Exposure Reduction
Classic, effective controls include:- Locating PI Web API instances and all associated control system devices behind robust firewalls.
- Ensuring no ICS device is reachable from the open internet.
- Isolating ICS networks from business networks to limit the blast radius of a compromise.
- Employing secure remote access solutions like up-to-date VPNs, but with recognition of their own vulnerabilities and the security hygiene of client devices.
5. Monitoring, Detection, and Social Engineering Awareness
- Monitor for anomalous uploads, privilege escalations, or suspicious annotation activity within PI Web API logs.
- Regularly educate users about social engineering tactics, phishing, and the criticality of never disabling security features.
- Leverage resources such as the CISA “Recognizing and Avoiding Email Scams” guide, along with broader ICS defense-in-depth documentation.
6. Incident Preparedness
Organizations should have clear, practiced internal procedures for reporting and investigating suspicious events. Any unusual behavior or suspected exploitation should also be reported to CISA for cross-industry analysis.Analysis: Weighing Notable Strengths and Persistent Risks
Strengths of Response
- Transparency: Both AVEVA and CISA have provided proactive, clear guidance—not only technical details but contextual mitigations specific to industrial environments.
- Layered Mitigation: The response emphasizes not just patching, but hardening operational practices—a recognition that technological fixes alone cannot close all windows of attack.
- No Evidence of Exploitation: As of June 2025, there have been no publicly reported attacks leveraging this vulnerability, lending credence to the assertion of high attack complexity and suggesting that prompt patching can prevent incidents.
Risks and Ongoing Weaknesses
- Human Factor Remains the Wildcard: The attack chain relies on successful social engineering—a technique that repeatedly outpaces technological defenses. In less mature organizations, the risk is magnified by limited user training and inconsistent policy enforcement.
- Brittle Legacy Systems: Production environments built atop legacy architectures struggle with timely patching, user management, and network segmentation. These gaps widen the window of vulnerability well beyond environments that promptly apply updates.
- Supply Chain Implications: In cases where integrators or third-party vendors have administrative or annotation privileges, risk may shift outside the direct control of asset owners—prompting the need for comprehensive supply chain security reviews.
Cautions and Unverified Vectors
While current reports flag the need for CSP disabling (which in itself is rare and would typically be cautioned against by IT), there is always the risk that further research will uncover bypasses or alternative exploit vectors, especially as attackers probe for related issues in the API’s content handling. Organizations are thus advised to monitor threat intelligence for updates and remain vigilant for related advisories.Broader Implications: Security in an Interconnected ICS World
The PI Web API incident is emblematic of broader ICS cybersecurity challenges:- As industrial organizations digitize and converge IT and OT, attack surfaces increase exponentially.
- Features like annotation attachments, designed for collaborative operational efficiency, paradoxically expand risk unless rigorously controlled.
- The complexity of patching and privilege management, compounded by the global diffusion of these systems, makes coordinated, ongoing security improvement a necessity.
Conclusion: Balancing Usability and Security in PI Web API Deployments
AVEVA’s PI Web API vulnerability serves as a timely reminder that convenience, collaboration, and accessibility in industrial data management always carry an attendant responsibility to secure. The sophisticated exploit path outlined by CVE-2025-2745 reinforces the notion that defense is a moving target, as adversaries blend technical and social engineering vectors in pursuit of access.For administrators and security professionals, the call to action is unmistakable:
- Patch immediately.
- Audit and minimize privileges.
- Harden browser and attachment handling.
- Educate every user, from operator to engineer, on the tactics adversaries may deploy.
Above all, the lesson is one of continuous learning and adaptation. Today’s mitigations are only as strong as tomorrow’s weakest link—and in the world of critical infrastructure, the cost of complacency is simply too high.
Source: CISA AVEVA PI Web API | CISA