AI responsibility at executive board level, but little information on its own AI models
Overall rating “solid”. No deduction for controversies. The superscript numbers link directly to the source.
1 · Responsibility
Infineon describes a group-wide AI framework with a control procedure, training and a Corporate AI Rule, but the rule is not public, the supervisory board does not explicitly address AI, and figures are missing.
What is documented
- The portfolio of the Chief Digital and Sustainability Officer covers artificial intelligence, digitalisation and data strategy1
- An internal control procedure applies to all AI users and AI development teams, with a Corporate AI Rule, minimum standards and training2
- According to the manifesto, a clearing function from engineering and compliance reviews AI projects and escalates to a Digital Council5
- The Integrity Line allows anonymous reports; the audit committee receives quarterly compliance reports8 2
- Infineon discloses its position on implementing the EU AI Act in the lobbying register10
The case against
Self-description without audit
- The text of the Corporate AI Rule and the risk classes and approval criteria for AI are not public
- According to the materiality analysis, AI is not a material sustainability topic, so AI figures are missing9
- No external audit or certification of AI management was found
Gaps in oversight and complaint channels
- The supervisory board's technology and digitalisation committee has no explicit AI responsibility; AI was not a meeting topic in 20251 4
- Whether the four AI agents in the HR service count as high-risk applications under the EU AI Act is not disclosed7
- No AI-specific complaint channel for affected persons is documented
Sources for Responsibility (10)
- Infineon Technologies AG: Annual Report 2025 · 27 November 2025 · primary source
- Infineon Technologies AG: Statement on Corporate Governance 2025 · 27 November 2025 · primary source
- 1E9 Festival: Speaker profile Christiane Miethge · 3 November 2026 · secondary source
- Infineon Technologies AG: Supervisory Board (committees and members) · 3 November 2026 · primary source
- Infineon Technologies AG: Responsible Artificial Intelligence (AI) at Infineon (manifesto) · 3 November 2026 · secondary source
- OutSystems: Infineon Low-Code Generative AI (case study) · 3 November 2026 · secondary source
- ServiceNow: Customer Session: How Infineon pioneered Agentic AI for HR (AI Summit Munich) · 7 October 2026 · secondary source
- Infineon Technologies AG: Sustainability at Infineon 2025 (Supplement to the Annual Report 2025) · 27 November 2025 · primary source
- Infineon Technologies AG: Sustainability at Infineon 2025 (German version) · 27 November 2025 · primary source
- Deutscher Bundestag, Lobbyregister: Regulatory project RV0024234 – implementation of the EU AI Act, entry Infineon Technologies AG · 7 May 2026 · primary source
2 · Safety
Infineon documents certified security management with its own security organisation, but the EUCLEAK vulnerability in its own cryptographic library weighs against this, and AI testing is available only in isolated cases.
What is documented
- Certificate IS 829901 to ISO/IEC 27001:2022 covers the central ISMS and data centre operations at three sites5
- The security organisation comprises a CISO, Cyber Defense Center, CERT and PSIRT with channels for reporting vulnerabilities6
- The audit committee receives reports on cyber risks, external audits, incidents and critical vulnerabilities7
- The fall detection test report states the method, a sensitivity of 93.3 % and the false alarm rate1
- A patch exists for EUCLEAK; the BSI recertified three patched TPMs8 4
The case against
Vulnerabilities
Sources for Safety (10)
- Imagimob, an Infineon Technologies company: DEEPCRAFT Ready Model for Fall Detection – Test Report · 3 November 2026 · primary source
- Electronics Era (Infineon press release): AURIX TC4x MCU for embedded AI application development receives safety assessment from Fraunhofer IKS · 10 April 2024 · secondary source
- Infineon Technologies AG: Responsible Artificial Intelligence (AI) at Infineon (manifesto) · 3 November 2026 · secondary source
- heise online: EUCLEAK: More products vulnerable to cloning attack · 9 September 2024 · primary source
- BSI Group (Zertifizierungsstelle): Client Directory Profile Infineon Technologies AG – ISO/IEC 27001:2022, certificate IS 829901 · 8 May 2026 · primary source
- Infineon Technologies AG: Infineon Cyber Security Program · 3 November 2026 · primary source
- Infineon Technologies AG: Statement on Corporate Governance 2025 · 27 November 2025 · primary source
- NinjaLab (Thomas Roche): EUCLEAK – Side-Channel Attack on the YubiKey 5 Series · 3 September 2024 · primary source
- Imagimob: Ready Models FAQ · 3 November 2026 · secondary source
- ServiceNow: Customer Session: How Infineon pioneered Agentic AI for HR (AI Summit Munich) · 7 October 2026 · secondary source
3 · Transparency
Infineon names its main internal AI applications and promises disclosure, but provides performance, data and limitation information only for the fall detection model.
What is documented
- The annual report names generative AI in customer support and AI in chip design, verification and HR1
- The manifesto announces disclosure of where AI is used and names explainability as a principle2
- The privacy policy names AI analysis of surveys and the use of chat content for training3
- The fall detection test report names limitations such as time lying after the fall, sensor position and a small sample4
The case against
Missing labelling
- There is no information on labelling AI-generated answers in customer support
- Whether employees are informed about the HR AI agents is not documented8
Sources for Transparency (8)
- Infineon Technologies AG: Annual Report 2025 · 27 November 2025 · primary source
- Infineon Technologies AG: Responsible Artificial Intelligence (AI) at Infineon (manifesto) · 3 November 2026 · secondary source
- Infineon Technologies AG: Privacy Policy · 3 November 2026 · secondary source
- Imagimob, an Infineon Technologies company: DEEPCRAFT Ready Model for Fall Detection – Test Report · 3 November 2026 · primary source
- Hackster.io: Infineon's Imagimob Edge AI Portfolio Is Now DEEPCRAFT, Including Newly-Launched Ready Models · 3 November 2026 · secondary source
- Electronics Era (Infineon press release): AURIX TC4x MCU for embedded AI application development receives safety assessment from Fraunhofer IKS · 10 April 2024 · secondary source
- Imagimob: Ready Models FAQ · 3 November 2026 · secondary source
- ServiceNow: Customer Session: How Infineon pioneered Agentic AI for HR (AI Summit Munich) · 7 October 2026 · secondary source
4 · Why the harm rule does not apply
The harm rule does not apply: within the 36-month window there is no judgment, no authority decision, no fine and no pattern of harm from its own AI. The two medium controversies concern hardware security and end use, with no AI link.
5 · Controversies (1)
- mediumResearchers found the EUCLEAK vulnerability in the cryptographic library of the security chips, which allows FIDO keys to be cloned; patched TPMs were recertified. · 3 September 2024 · abgeschlossen
6 · What remains open
- Text and implementation status of the Corporate AI Rule
- Explicit AI oversight in the supervisory board
- Cyber risks and incidents according to the 2025 risk and opportunity report
- Model information and usage restrictions for the edge AI models
7 · How it was assessed
Cut-off date: 3 November 2026 · 41 searches, 32 fetches.
Primary documents read (7)
- Annual Report 2025, pp. 1–40 (S03, two retrievals)
- Statement on Corporate Governance 2025 (S02, two retrievals)
- Sustainability at Infineon 2025, pp. 1–38 (S04, two retrievals) and Sustainability at Infineon 2025, pp. 1–22 (S05)
- HGB annual financial statements 2025 (S06), AGM voting results 2026 (S07)
- Responsible AI statement of principles (S01), Supervisory Board page (S08), Cyber Security Programme (S10), Privacy Policy (S20)
- Ad hoc announcement on Qimonda settlement (S29), fall detection test report (S16)
- Independent findings: BSI certificate register (S09), NinjaLab EUCLEAK (S11), Bundestag lobbying register (S21), Violation Tracker (S30)
Research of 3 November 2026 under Globalance AI Footprint rulebook version 0