Skip to content
GLOBALANCE INTELLIGENCEThe economy ahead.
This Week Index 28/100 Sign in
Money & Investments › GI 100 Index
The Great Value ShiftInvestment ThemesGI 100 IndexGI PortfolioStart Future Check
AI FOOTPRINTlimitedASSESSMENT 2026GLOBALANCEINTELLIGENCE
Globalance AI Footprint

Assessment as of November 2026 · next assessment 2027.

Result of the assessment by the Globalance research team under the published methodology. Not a certification or quality label.

How we assess ›

How this rating is builtVideo · 0:52
GI 100 Index · Profile

Johnson Controls International

IrelandEnablersAssignment FullData quality Medium

Key messagelimited AI Footprint

Johnson Controls uses artificial intelligence in building control and security technology and has explicitly assigned oversight of it to a board committee; however, the company has not published documented binding internal rules, review or approval procedures for AI. The protection of its products against cyber attacks is documented by independent certificates and orderly security advisories; public information on testing of the AI functions, on their limitations and on the promised model descriptions is missing, and a lawsuit is pending over a data breach in 2023.

Opportunities

  • Safetysolid

    A certified secure development process under IEC 62443-4-1 (ISASecure SDLA, expired 2026-09-01, renewal not verified) and a component certification under IEC 62443-4-2 are independently documented (S30, S32).

    A certified secure development process under IEC 62443-4-1 (ISASecure SDLA, expired 2026-09-01, renewal not verified) and a component certification under IEC 62443-4-2 are independently documented (S30, S32). For system protection, a working vulnerability process with Johnson Controls as a CVE numbering authority and advisories published by CISA, as well as ISO 27001 and SOC 2 information for OpenBlue, are documented (S18, S19, S23). This evidence is not AI-specific; for testing, validation or red-teaming of the autonomous and agentic AI functions and of facial recognition, there are only principles and product statements (S01, S15, S24). The pending class action over the 2023 data breach, in which those affected were informed only around 21 months after discovery, is recorded as a medium controversy without AI connection (S10, S12).

Risks

  • Responsibilitylimited

    Condition (a) of the minimum requirement is met: the charter of the Governance and Sustainability Committee explicitly names artificial intelligence under technology risk, and the 2026 proxy statement reports that the committee expanded its charter to cover AI oversight and worked in depth with …

    Condition (a) of the minimum requirement is met: the charter of the Governance and Sustainability Committee explicitly names artificial intelligence under technology risk, and the 2026 proxy statement reports that the committee expanded its charter to cover AI oversight and worked in depth with management on AI in 2025 (S06, S05, S04). Condition (b) is not met: AI ethics principles with an “Enterprise AI Council” that is not described further, AI Terms and an AI Code of Conduct with prohibited uses for customers are available only as lower-grade evidence (S01, S27, S31), and no internal AI policy, risk classification or approval procedure with primary evidence was found. The pillar is therefore “limited” on the basis of principles, without a documented material weakness; AI figures, AI training and an AI-specific complaints channel are also missing.

  • Transparencylimited

    The AI principles promise disclosure of purpose and data types, information on limitations and error rates, and model cards for AI affecting rights or safety (S01); no such model cards or error rates were found.

    The AI principles promise disclosure of purpose and data types, information on limitations and error rates, and model cards for AI affecting rights or safety (S01); no such model cards or error rates were found. The AI Terms state that outputs may be inaccurate and that customer content is not used for other customers' models, but name no models used (S31). The OpenBlue product sheet discloses hosting and sub-processors, but not the use of data for AI (S23). It remains at principles without evidence of implementation.

  • Risk area

    Facial recognition and video analytics in security products, including for authorities

  • Risk area

    Autonomous and agentic control of building technology

  • No serious controversy recorded.

What lies behind the assessment

Research of 3 November 2026
44searches
45pages and documents read
12primary documents
2controversies checked
6open points, named openly
The full research: pillars, evidence, controversies and open points
AI Footprint · Enabler · As of 3 November 2026
Johnson Controls InternationalOverall rating: limited

Johnson Controls has assigned oversight of AI to a board committee, but documents no binding internal AI rules. Product cybersecurity is certified, but testing of the AI functions and the promised model cards are missing.

Responsibility

40 %
limitedsolidstrong

A board committee oversees AI, but binding AI rules are not documented.

  • +Committee charter explicitly names AI under technology risk Source ↗
  • +Proxy 2026: charter expanded to cover AI oversight Source ↗
  • –Enterprise AI Council without mandate, composition or reports Source ↗
  • –No internal AI policy, risk classes or approval procedure
  • –No AI training and no AI figures documented
All sources for this pillar (12)

Safety

40 %
limitedsolidstrong

Independent certificates for product security, but not specific to AI.

  • +ISASecure certificate for the secure development process Source ↗
  • +Orderly vulnerability advisories with patches and CISA advisories Source ↗
  • +ISO 27001 and SOC 2 Type II for OpenBlue Enterprise Manager Source ↗
  • –No testing of the autonomous and agentic AI functions documented
  • –Class action over the 2023 data breach with late notification Source ↗
All sources for this pillar (21)

Transparency

20 %
limitedsolidstrong

Model cards are promised but not published.

  • +Principles promise information on purpose, limitations and error rates Source ↗
  • +AI Terms: customer content not used for other customers' models Source ↗
  • –Promised model cards not found
  • –Foundation models used are not named Source ↗
All sources for this pillar (8)
What is missing for “strong”

These steps are in the company’s own hands.

1Responsibility
Disclose AI policy and AI Council

Publish an internal AI policy with risk classes and document the mandate and work of the Enterprise AI Council.

2Safety
Document testing of the AI functions

Disclose testing, validation and red-teaming of the autonomous OpenBlue functions and facial recognition.

3Transparency
Publish the promised model cards

Publish model cards with limitations, error rates and foundation models used for the main AI products.

Context: 101 companies assessed

Distribution of overall ratings, no ranking
strong
none yet
solid
limited
← Johnson Controls International is one of 66

No company reaches “strong” today. Disclosure on how companies handle AI is still young across the market.

Controversies
2 1 medium, 1 minor

No deduction for controversies.

Largest risk area
Malfunction of autonomous building control

For testing, validation or red-teaming of the autonomous and agentic AI functions in OpenBlue, there are only principles, no evidence of testing.

How thoroughly assessed
41 Evidence

12 primary documents read, 44 searches.

Read the full reasoning and all sources
Johnson Controls International · Full reasoning

AI oversight at board level, but no documented AI rules and no testing of the AI functions

Overall rating “limited”. No deduction for controversies. The superscript numbers link directly to the source.

1 · Responsibility

limitedWeight 40%

AI oversight at board level is documented, but rules on reviewing and approving AI exist only as principles and contract terms without evidence of implementation.

What is documented
  • The charter of the Governance and Sustainability Committee of September 2025 explicitly names AI under technology risk1
  • According to the 2026 proxy, the committee expanded its charter to cover AI oversight and worked in depth with management on AI in 20253 2
  • AI is a topic of the quarterly cyber reports by the CDIO and CISO to the board3
  • AI ethics principles, AI Terms and an AI Code of Conduct govern development and customer use5 6 7
The case against
Principles without evidence of implementation
  • The Enterprise AI Council is mentioned without mandate, composition or reporting lines5
  • The regular audits promised in the principles are not documented with results5
  • No internal AI policy, risk classes or approval for in-house and acquired AI were found
No AI-specific embedding
  • The code of ethics contains no section on AI; the integrity helpline is not AI-specific10 9
  • No AI training, AI figures or external AI audits were found
Sources for Responsibility (12)
  1. Johnson Controls International plc: Governance and Sustainability Committee Charter · 10 September 2025 · primary source
  2. Johnson Controls International plc / U.S. SEC: Form ARS FY2025 (includes 2026 proxy statement) · 16 January 2026 · primary source
  3. Johnson Controls International plc / U.S. SEC: Notice of 2026 Annual General Meeting and Proxy Statement (DEF 14A) · 16 January 2026 · primary source
  4. Johnson Controls: Two Johnson Controls leaders named to Constellation Research Artificial Intelligence 150 list · 12 June 2025 · secondary source
  5. Johnson Controls: Artificial Intelligence: Trust in Johnson Controls (AI Ethics Principles) · 3 November 2026 · secondary source
  6. Johnson Controls: AI Code of Conduct · 5 November 2025 · secondary source
  7. Johnson Controls: AI Terms · 3 November 2026 · secondary source
  8. Johnson Controls International plc / U.S. SEC: Form 10-K for the fiscal year ended 2025-09-30 · 14 November 2025 · primary source
  9. Johnson Controls International plc: Irish Statutory Reports and Non-Financial Report for Fiscal 2025 · 3 November 2026 · secondary source
  10. Johnson Controls International plc: 2024 JCI Code of Ethics · 3 November 2026 · secondary source
  11. Johnson Controls: Ethisphere names Johnson Controls as one of the World's Most Ethical Companies, marking 19 years on the list · 18 March 2026 · secondary source
  12. Johnson Controls: 2025 Sustainability Report – GRI Index · 3 November 2026 · secondary source

2 · Safety

solidWeight 40%

Johnson Controls independently documents secure product development and vulnerability advisories, but testing of the AI functions is missing and a lawsuit over the 2023 data breach is pending.

What is documented
  • An ISASecure SDLA certificate documents the secure development process under IEC 62443-4-1; it expired on 2026-09-011
  • A component certification under IEC 62443-4-2 exists for a chiller control system2
  • Johnson Controls is a CVE numbering authority; advisories on iSTAR and Metasys were issued with patches and by CISA7 8
  • ISO 27001, SOC 2 Type II and AES-256 encryption are stated for OpenBlue Enterprise Manager9
  • The AI Code of Conduct prohibits 23 uses, such as real-time remote identification by law enforcement15
The case against
No AI-specific evidence of testing
  • For testing, validation and red-teaming of OpenBlue AI, OBI and Nantum AI there are only principles3
  • Error rates and bias testing of facial recognition are not published
  • No renewal of the SDLA certificate after its expiry on 2026-09-01 was found1
2023 data breach
  • Access from February to September 2023 went undetected; those affected were notified only from June 202512 13
  • According to the class action, protective measures and training were inadequate; the status of the case is open21
Sources for Safety (21)
  1. ISASecure / exida: ISASecure SDLA 3.0.0 Certificate JCI 2304032 C001 (Global Secure Product Development Process) · 14 August 2023 · primary source
  2. ISASecure / exida: ISASecure CSA Certificate JCI 2110089 C002 (YMC2 ModB Centrifugal Chiller, OptiView) · 15 May 2025 · primary source
  3. Johnson Controls: Artificial Intelligence: Trust in Johnson Controls (AI Ethics Principles) · 3 November 2026 · secondary source
  4. Johnson Controls: Johnson Controls unveils AI-powered OpenBlue platform expansions · 5 August 2026 · secondary source
  5. Johnson Controls: Johnson Controls acquires Nantum AI to accelerate AI-driven energy optimisation · 27 April 2026 · secondary source
  6. Exacq Technologies (Johnson Controls): Tyco AI (product page, staging domain) · 3 November 2026 · secondary source
  7. GCVE / CIRCL (database entry for CISA ICSA-26-022-04): ICSA-26-022-04 Johnson Controls iSTAR Configuration Utility (CVE-2025-26386) · 22 January 2026 · primary source
  8. CVE Program (MITRE, CVE Services API): CVE-2025-26385 Metasys command injection · 30 January 2026 · primary source
  9. Johnson Controls: OpenBlue Enterprise Manager – Cybersecurity and Data Privacy Sheet (GPS0050-CE-EN Rev B) · 16 December 2024 · primary source
  10. board-cybersecurity.com: Johnson Controls International plc 10-K Item 1C. Cybersecurity – 2025-11-14 · 14 November 2025 · secondary source
  11. Johnson Controls: Trust Centre – Cybersecurity · 3 November 2026 · secondary source
  12. Johnson Controls / Office of the Attorney General of California: Notice of Data Breach (sample letter filed with the California Attorney General) · 30 June 2025 · primary source
  13. Iowa Attorney General: Johnson Controls – data breach notice of 2025-06-30 · 30 June 2025 · primary source
  14. Johnson Controls International plc / U.S. SEC: Form 10-K for the fiscal year ended 2025-09-30 · 14 November 2025 · primary source
  15. Johnson Controls: AI Code of Conduct · 5 November 2025 · secondary source
  16. Johnson Controls: AI Terms · 3 November 2026 · secondary source
  17. Johnson Controls (Security Products): Comprehensive Solutions to Secure our National Buildings (Federal Government) · 3 November 2026 · secondary source
  18. Commission de la protection de la vie privée (Belgien): Opinion no. 26/2015 on the Binding Corporate Rules of Johnson Controls · 1 July 2015 · primary source
  19. Johnson Controls: International Data Transfers · 3 November 2026 · secondary source
  20. Johnson Controls: Trust Centre – Privacy · 3 November 2026 · secondary source
  21. BizTimes Milwaukee: Johnson Controls faces class action lawsuit following 2023 data breach · 7 July 2025 · secondary source

3 · Transparency

limitedWeight 20%

Johnson Controls promises information on limitations, error rates and model cards, but none of it was found, and the models used are not named.

What is documented
  • The AI principles commit to making clear when, why and how AI is used1
  • The AI Terms state that outputs may be inaccurate and that customer content is not used for other customers' models5
  • The OpenBlue product sheet discloses hosting and sub-processors7
The case against
Commitments without implementation
  • The promised model cards for AI affecting rights or safety were not found despite a targeted search1
  • Documented limitations and error rates for OpenBlue AI and face matching are missing
Missing model information
  • Foundation models used, third-party providers and training data for facial recognition are not named5
  • It is not disclosed per product which functions use AI and with what degree of autonomy
Sources for Transparency (8)
  1. Johnson Controls: Artificial Intelligence: Trust in Johnson Controls (AI Ethics Principles) · 3 November 2026 · secondary source
  2. Johnson Controls: AI Code of Conduct · 5 November 2025 · secondary source
  3. Johnson Controls International plc / U.S. SEC: Form 10-K for the fiscal year ended 2025-09-30 · 14 November 2025 · primary source
  4. Johnson Controls: Legal – Digital Terms (overview) · 3 November 2026 · secondary source
  5. Johnson Controls: AI Terms · 3 November 2026 · secondary source
  6. Johnson Controls International plc: Irish Statutory Reports and Non-Financial Report for Fiscal 2025 · 3 November 2026 · secondary source
  7. Johnson Controls: OpenBlue Enterprise Manager – Cybersecurity and Data Privacy Sheet (GPS0050-CE-EN Rev B) · 16 December 2024 · primary source
  8. Memoori Research AB: Johnson Controls AI Strategy in Smart Buildings: An Independent Assessment · 2 May 2026 · secondary source

4 · Why the harm rule does not apply

The harm rule does not apply: within the 36-month window, neither an AI-related judgment nor a fine nor a pattern of serious incidents was found.

5 · Controversies (2)

6 · What remains open

  • Internal AI policy, risk classification and approval procedure for in-house and acquired AI
  • Mandate and work of the Enterprise AI Council
  • Testing and red-teaming of the autonomous and agentic OpenBlue functions, error rates of facial recognition
  • How the prohibited uses are enforced when facial recognition is sold to authorities
  • Sustainability reports 2025 and 2026, which could not be retrieved
  • Renewal of the ISASecure SDLA certificate after 2026-09-01

7 · How it was assessed

Cut-off date: 3 November 2026 · 44 searches, 45 fetches.

Primary documents read (12)
  • Form 10-K FY2025 (risk factors, up to cut-off)
  • Proxy Statement 2026 (DEF 14A) and Form ARS FY2025, including section Oversight of Cybersecurity (p. 29)
  • Governance and Sustainability Committee Charter dated 2025-09-10
  • Irish Statutory Reports and Non-Financial Report for Fiscal 2025 (complete, with read-more)
  • Form 10-Q Q3 FY2026 (notes, up to cut-off)
  • Form 8-K Annual General Meeting 2025
  • AI Ethics Principles, AI Terms, AI Code of Conduct, Code of Ethics 2024
  • OpenBlue Enterprise Manager Cybersecurity and Data Privacy Sheet
  • ISASecure certificates SDLA (JCI 2304032) and CSA (JCI 2110089)
  • Data breach notifications filed with the California and Iowa Attorneys General
  • CVE record CVE-2025-26385 and advisory ICSA-26-022-04
  • Belgian BCR opinion Avis n° 26/2015

Research of 3 November 2026 under Globalance AI Footprint rulebook version 0

We assess documented practice in dealing with AI; this is not an investment recommendation. Three pillars, weights 40/40/20, harm rule.

Research of 3 November 2026 under Globalance AI Footprint rulebook version 0. Documented observations from public sources, no investment recommendation.

Important Notice. This material is provided for information purposes only and does not constitute an offer or solicitation to buy or sell any financial instrument, nor does it constitute investment advice or a recommendation. The GI 100 Index is a rules-based strategy index of Globalance Bank. Investments involve risk, including the loss of capital invested. The final product documents (including FinSA) shall prevail in all cases.

Your company? Information on inclusion in the GI 100 Index →

What comes next