AI oversight in the Risk Committee, but no AI policy and little disclosure
Overall rating “limited”. No deduction for controversies. The superscript numbers link directly to the source.
1 · Responsibility
Responsibility for AI risks is set out at board level, but UPS publishes neither an AI policy nor an approval or review procedure for AI.
What is documented
- The Risk Committee, made up solely of independent members, has been responsible for AI risks since 2025; its charter was revised for this purpose1
- The Chief Digital and Technology Officer reports to the committee regularly; the Chief Legal and Compliance Officer and the CISO also attend meetings1 2
- The collective agreement with the Teamsters bans inward-facing cameras and disciplinary action based solely on technology5
- Under the collective agreement, technological changes must be announced 45 days in advance5
- A general ethics hotline allows anonymous reports and protects against retaliation4
The case against
Rules and evidence
- No AI policy, risk classification or approval procedure for in-house and purchased AI was found; the code of conduct does not mention AI4
- The retrieved 2023 version of the Risk Committee charter does not yet mention AI; the revised version was not available3
- There are no AI metrics, no external audit and no certification of AI management
Monitoring of drivers
- In Sweden more than 50 employees reported an AI camera system to the data protection authority; according to the union, drivers had already been classed as “safe” or “unsafe”7
- An activist publication alleges that the Lytx sensors in US vehicles circumvent the ban on inward-facing cameras; this has not been independently confirmed6
Sources for Responsibility (7)
- United Parcel Service, Inc. (SEC filing): Notice of 2026 Annual Meeting and Proxy Statement (DEF 14A) · 19 March 2026 · primary source
- United Parcel Service, Inc. (SEC-Einreichung): Form 10-K for fiscal year 2025 · 17 February 2026 · primary source
- United Parcel Service, Inc.: Risk Committee Charter · 2 November 2023 · primary source
- United Parcel Service, Inc.: Code of Business Conduct and Ethics · 3 November 2026 · secondary source
- International Brotherhood of Teamsters: UPS National Master Agreement – Highlights (2023–2028) · 1 August 2023 · primary source
- World Socialist Web Site: Article on Lytx DriveCam systems in UPS vehicles (qtfu-d29) · 28 December 2025 · secondary source
- Sveriges Radio: UPS staff in Sweden report new AI camera system to the privacy authority · 3 November 2026 · primary source
2 · Safety
UPS describes a concrete security and incident regime in its 10-K, but testing of the AI systems themselves is documented only in trade press.
What is documented
- The 10-K describes an incident response plan with escalation up to the board, scenario exercises, internal audits and reviews by independent third parties5 6
- An Information Security & Privacy Governance Council is in place; suppliers are subject to reviews and contractual rules on data access5 6
- The risk factors in the 10-K expressly name the compromise of AI systems as a cyber risk5 7
- The privacy notice gives 2025 processing figures for data subject rights and rules out selling personal data to train third-party language models8
- According to the company, staff confirm every reply suggested by MeRA; the rollout was staged1
The case against
Testing of the AI systems
Protection against misuse
- The anti-fraud page warns about phishing but names no protection against AI-enabled identity misuse9
- A named security framework and AI-specific safeguards for models and training data are missing
Sources for Safety (10)
- CIO (Foundry): UPS delivers customer wins with generative AI · 3 May 2024 · secondary source
- Multichannel Merchant: UPS Capital Releases AI Tool to Combat Porch Piracy · 3 November 2026 · secondary source
- Supply Chain Dive: UPS AI employee upskilling and network changes (part 2 of a series on AI at parcel carriers) · 6 April 2026 · secondary source
- PYMNTS: UPS Automates 90% of Daily Customs Clearances With AI Agents · 30 September 2025 · secondary source
- United Parcel Service, Inc. (SEC-Einreichung): Form 10-K for fiscal year 2025 · 17 February 2026 · primary source
- board-cybersecurity.com: UNITED PARCEL SERVICE INC 10-K Item 1C. Cybersecurity – 2026-02-17 · 17 February 2026 · secondary source
- United Parcel Service, Inc. (SEC-Einreichung): Form 10-K for fiscal year 2024 · 3 November 2026 · primary source
- United Parcel Service, Inc.: UPS Privacy Notice (USA) with U.S. Addendum · 30 June 2026 · primary source
- United Parcel Service, Inc.: Protect Against Fraud · 7 October 2025 · secondary source
- United Parcel Service, Inc.: Code of Business Conduct and Ethics · 3 November 2026 · secondary source
3 · Transparency
AI use is disclosed only in general terms; explanations of individual systems come from trade press.
What is documented
- The privacy notice discloses the use of AI technologies, a third-party chatbot and the improvement of large language models1
- The risk factors in the 10-K cover AI use in general terms2
- Trade press describes the rating bands and data basis of DeliveryDefense: around 11 billion deliveries to around 130 million addresses5 4
The case against
System information missing
- There are no system or model descriptions for DeliveryDefense, MeRA or the customs AI and no AI transparency report
- Affected recipients are not told how their address is rated or how they can object4 5
- Labelling of AI-generated email and chatbot replies to customers is not documented3
- The GRI report classes customer privacy as not material6
Sources for Transparency (6)
- United Parcel Service, Inc.: UPS Privacy Notice (USA) with U.S. Addendum · 30 June 2026 · primary source
- United Parcel Service, Inc. (SEC-Einreichung): Form 10-K for fiscal year 2025 · 17 February 2026 · primary source
- CIO (Foundry): UPS delivers customer wins with generative AI · 3 May 2024 · secondary source
- CNBC: AI is policing package theft for UPS as porch piracy surge continues · 21 September 2023 · secondary source
- Multichannel Merchant: UPS Capital Releases AI Tool to Combat Porch Piracy · 3 November 2026 · secondary source
- United Parcel Service, Inc.: 2025 UPS GRI Report · 13 March 2026 · primary source
4 · Why the harm rule does not apply
The harm rule does not apply: no ruling, regulatory decision or pattern of serious AI incidents was found in the 36-month window. A decision by the Swedish data protection authority could change this.
5 · Controversies (2)
- mediumMore than 50 employees reported an AI camera system monitoring driving behaviour in UPS vehicles in Sweden to the data protection authority; the outcome is unknown. · · unclear
- minorAccording to an activist publication, Lytx DriveCam sensors in US vehicles circumvent the collective agreement's ban on inward-facing cameras; this is not independently confirmed. · 28 December 2025 · unklar
6 · What remains open
- Date, status and outcome of the report of the AI camera system to the Swedish data protection authority
- Revised Risk Committee charter with AI oversight; only the 2023 version was retrieved
- AI policy, responsible AI report and system description for DeliveryDefense were not found
- Investor exclusions and military and government use were not researched because the search budget was exhausted
- Approvals and safety evidence for drone and autonomous logistics
- External information security certifications at group level
7 · How it was assessed
Cut-off date: 3 November 2026 · 36 searches, 23 fetches.
Primary documents read (9)
- Form 10-K FY 2025 (S1, with continued reading from character 100000)
- Form 10-K FY 2024 (S3, characters 0–100000)
- Proxy Statement 2026 DEF 14A (S4, with continued reading from character 100000)
- Risk Committee Charter (S5)
- 2025 UPS GRI Report (S6)
- Code of Business Conduct and Ethics (S9)
- UPS Privacy Notice USA (S10)
- UPS Protect Against Fraud (S11)
- UPS-Teamsters National Master Agreement Highlights (S17)
Research of 3 November 2026 under Globalance AI Footprint rulebook version 0