Board-approved AI policy, but little information on its own AI product
Overall rating “solid”. No deduction for controversies. The superscript numbers link directly to the source.
1 · Responsibility
Prysmian governs AI through a board-approved policy with roles, a register and risk classes, but an approval process, AI metrics and committee responsibility are missing.
What is documented
- The AI policy approved by the board on 2025-10-29 defines roles for the AI Governance Committee, DPO, cybersecurity, procurement and the CIO office with an AI inventory1 2
- The policy requires risk classification under the EU AI Act, data protection impact assessments for high-risk systems and contract clauses in procurement1 2
- Unapproved AI plug-ins and public tools with company accounts are prohibited; data uploads to third-party AI require approval1
- The Information Security & AI Committee under the Group Chief Information & Digital Officer also includes Risk & Compliance and Internal Audit3
- Ethical concerns about AI can be reported to the DPO or via the helpline1 10
The case against
Gaps in implementation
Sources for Responsibility (11)
- Prysmian Group (na.prysmian.com): Prysmian AI Policy (AI Governance – AI Policy, English version) · 29 October 2025 · primary source
- Prysmian S.p.A.: Prysmian – AI Policy V4.2 (Rev. ITA) · 29 October 2025 · primary source
- Prysmian S.p.A.: Cybersecurity, Information Security & AI Governance (English website) · 3 November 2026 · secondary source
- Prysmian S.p.A.: AI program (Italian website «Innovazione – AI program») · 3 November 2026 · secondary source
- Prysmian S.p.A.: Corporate Governance Regulation (2026 version) · 25 February 2026 · primary source
- Prysmian S.p.A.: Committees (website) · 3 November 2026 · secondary source
- Prysmian S.p.A.: Report on Corporate Governance and Ownership Structure (financial year 2025) · 25 February 2026 · secondary source
- AI Governance Library (aigl.blog), Jakub Szarmach: Review: Prysmian – AI Governance – AI Policy · 31 August 2026 · secondary source
- Prysmian S.p.A.: Modello Organizzativo Privacy (ITA) · 7 May 2025 · primary source
- Prysmian S.p.A.: Prysmian Helpline System (whistleblowing website) · 3 November 2026 · secondary source
- Prysmian S.p.A.: Integrated Annual Report 2025 («Accelerating Growth») · 3 November 2026 · secondary source
2 · Safety
Prysmian documents a security system certified to ISO/IEC 27001 with approved policies and metrics, but no test results for the AI product PRY-CAM BRAIN are available.
What is documented
- RINA certified the information and cybersecurity services of Prysmian S.p.A. and its controlled companies to ISO/IEC 27001:2022, valid until 2027-12-195
- Board-approved policies govern information security, incident management, IoT/OT security with segmentation, and third parties12 6 7 8
- In 2024, 89% of log sources were captured in the SIEM; high-risk vulnerabilities were remediated within 13 weeks9
- The AI policy requires human oversight, verification of outputs and independent testing and attack simulations for high-risk systems1 2
- The AI policy prohibits manipulative, biometric categorisation and emotion recognition AI and requires AI malfunctions to be reported to the DPO2
The case against
Missing evidence of AI testing
- On validation, error rates and thresholds of PRY-CAM BRAIN there are only marketing claims without figures
- Test and approval records for internal AI such as pricing, contract review and the chatbot are not published
- No public channel for reporting product vulnerabilities was found
Sources for Safety (14)
- Prysmian Group (na.prysmian.com): Prysmian AI Policy (AI Governance – AI Policy, English version) · 29 October 2025 · primary source
- Prysmian S.p.A.: Prysmian – AI Policy V4.2 (Rev. ITA) · 29 October 2025 · primary source
- Prysmian S.p.A.: Fixed Monitoring Solution (PRY-CAM, product page) · 3 November 2026 · secondary source
- Prysmian S.p.A.: PRY-CAM, una rivoluzione nella gestione delle reti elettriche (story) · 3 November 2026 · secondary source
- RINA Services S.p.A.: ISO/IEC 27001:2022 certificate for Prysmian S.p.A. (no. 978/24) · 31 October 2024 · primary source
- Prysmian S.p.A.: Information Security Policy (PO-IT-I&C-SEC001) · 9 April 2025 · primary source
- Prysmian S.p.A.: IoT/OT Security Policy (PO-IT-I&C-SEC016) · 10 June 2025 · primary source
- Prysmian S.p.A.: Third Parties Security Policy (PO-IT-I&C-SEC012) · 10 June 2025 · primary source
- Prysmian S.p.A.: Integrated Annual Report 2024 – Performance ESG (online report) · 3 November 2026 · primary source
- Prysmian S.p.A.: Assemblea Prysmian S.p.A. 16-04-2026 – Risposte a domande preassembleari · 16 April 2026 · secondary source
- Fortinet: Customer reference Prysmian Group · 3 November 2026 · secondary source
- Prysmian S.p.A.: Security Incident Management Policy (PO-IT-I&C-SEC002) · 10 June 2025 · primary source
- Prysmian S.p.A.: Modello Organizzativo Privacy (ITA) · 7 May 2025 · primary source
- Prysmian S.p.A.: Cybersecurity, Information Security & AI Governance (English website) · 3 November 2026 · secondary source
3 · Transparency
Prysmian publishes its AI policy, the responsible body and five internal use cases, but discloses nothing verifiable on the limits and functioning of PRY-CAM BRAIN.
What is documented
- The AI policy and a website on AI governance with the composition of the responsible body are public1 3
- Employees must attribute AI outputs to the tool and label artificially generated image, audio and video content1
- The Digital & AI Program describes five use cases from OEE optimisation to contract review4 5
The case against
No product information
- For PRY-CAM BRAIN there are no documented limits, false alarm rates or explanations of the diagnosis
- The AI policy assigns technical documentation to the CIO office, but publishes no content2
- Which language models the «Virtual Expert» chatbot uses is not stated
Sources for Transparency (7)
- Prysmian Group (na.prysmian.com): Prysmian AI Policy (AI Governance – AI Policy, English version) · 29 October 2025 · primary source
- Prysmian S.p.A.: Prysmian – AI Policy V4.2 (Rev. ITA) · 29 October 2025 · primary source
- Prysmian S.p.A.: Cybersecurity, Information Security & AI Governance (English website) · 3 November 2026 · secondary source
- Prysmian S.p.A.: AI program (Italian website «Innovazione – AI program») · 3 November 2026 · secondary source
- Prysmian S.p.A.: AI program (English website) · 3 November 2026 · secondary source
- Prysmian S.p.A.: Fixed Monitoring Solution (PRY-CAM, product page) · 3 November 2026 · secondary source
- Prysmian S.p.A.: PRY-CAM, una rivoluzione nella gestione delle reti elettriche (story) · 3 November 2026 · secondary source
4 · Why the harm rule does not apply
The harm rule does not apply: there is no controversy with an AI, data or cyber connection within the window.
5 · Controversies (0)
No AI controversy found.
6 · What remains open
- AI responsibility of a board committee and reporting line of the Information Security & AI Committee
- Approval process and risk classification of specific AI systems such as PRY-CAM BRAIN and SAP Business AI
- AI metrics and AI disclosures in the 2025 sustainability statement, which was truncated on retrieval
- Validation, limits and system documentation of PRY-CAM BRAIN
7 · How it was assessed
Cut-off date: 3 November 2026 · 41 searches, 45 fetches.
Primary documents read (11)
- Integrated Annual Report 2025 (pp. I–34 and pp. 326–356 across two copies)
- Report on Corporate Governance and Ownership Structure, financial year 2025 (pp. 1–38)
- Corporate Governance Regulation, version dated 25 February 2026
- AI Policy (English and Italian, V4.2), approved 29 October 2025
- Information Security Policy, Security Incident Management Policy, IoT/OT Security Policy, Third Parties Security Policy
- Modello Organizzativo Privacy (7 May 2025)
- ISO/IEC 27001:2022-Zertifikat RINA Nr. 978/24
- Responses to questions ahead of the AGM of 16 April 2026 and voting results
- Integrated Annual Report 2024, online ESG key figures
- Third Quarter Financial Report 2026 (pp. 1–50)
- DOJ press release of 23 July 2024 and AGCM press release I873 of 12 December 2024
Research of 3 November 2026 under Globalance AI Footprint rulebook version 0