If you develop AI systems for battery management, state-of-health estimation, thermal runaway detection, or second-life battery assessment placed on the EU market, Article 110 of the EU AI Act directly governs your compliance obligations. EU AI Act Art.110 amends Directive 2006/66/EC — the EU Batteries Directive — to formally integrate EU AI Act requirements into the battery conformity framework, extending the high-risk AI pathway to AI-enabled battery management systems, state-of-health estimators, and safety-critical battery analytics platforms.

The amendment follows the same Annex I bridge mechanism as Art.104 (machinery), Art.105 (agricultural vehicles), Art.106 (two/three-wheel vehicles), Art.107 (motor vehicles), Art.108 (marine equipment), and Art.109 (radio equipment): AI systems embedded in battery products that qualify as safety components, health monitoring systems, or protection controllers under Directive 2006/66/EC now sit within EU AI Act Annex I, activating the Art.6(1) high-risk pathway and triggering the full Title III EU AI Act compliance stack. For battery AI developers and OEMs, this creates dual obligations spanning both batteries directive conformity and EU AI Act technical documentation, QMS, conformity assessment, and post-market monitoring.

Art.110 introduces one compliance complexity absent from the other sector amendments in the Art.104–112 series: the directive amended by Art.110 — Directive 2006/66/EC — has itself been substantially superseded by Regulation (EU) 2023/1542, the new Batteries Regulation, which entered into force in August 2023 and applies in phased stages from February 2024. Understanding how Art.110 interacts with this regulatory transition is a prerequisite for compliance planning.

What Directive 2006/66/EC Covers

Directive 2006/66/EC on batteries and accumulators governed the placing on the EU market of all batteries across three categories: portable batteries (consumer electronics, power tools), automotive batteries (starter/lighting/ignition for conventional vehicles), and industrial batteries (forklift trucks, grid storage, telecoms backup). The directive established minimum requirements for battery design, labelling, collection, recycling, and hazardous substance restrictions (cadmium, mercury, lead limits).

The battery categories covered by Directive 2006/66/EC and their AI relevance are:

The Directive 2006/66/EC to Regulation 2023/1542 Transition

Art.110 amends Directive 2006/66/EC. However, Regulation (EU) 2023/1542 — the new Batteries Regulation — repeals and replaces Directive 2006/66/EC in stages. Understanding the transition timeline is essential for compliance planning:

August 2023: Regulation (EU) 2023/1542 entered into force. Directive 2006/66/EC remained applicable during the transition period.

February 2024: The new Batteries Regulation began its phased application, with portable battery sustainability and labelling requirements becoming applicable for certain categories.

2025–2027: Progressive application of EV battery, industrial battery, and Battery Digital Passport requirements, with full repeal of Directive 2006/66/EC scheduled as each category-specific provision becomes applicable under 2023/1542.

For EU AI Act Art.110 compliance, this transition creates a practical ambiguity: the Annex I bridge inserted by Art.110 references Directive 2006/66/EC, but as the directive phases out and Regulation 2023/1542 phases in, the operative product regulation governing battery conformity changes. In practice:

• During the transition period: Battery AI systems covered by Directive 2006/66/EC fall within Art.110's Annex I bridge and are subject to the EU AI Act high-risk pathway.
• After full application of Regulation 2023/1542: The EU AI Act's Annex I cross-reference mechanism encompasses successor regulations. Battery AI systems within Regulation 2023/1542's scope — particularly EV traction batteries, industrial batteries, and stationary storage batteries — remain within the high-risk pathway through the same Annex I functional logic even as the underlying product regulation changes.
• Battery Digital Passport (BDP): Regulation 2023/1542 mandates a Battery Digital Passport for EV batteries, industrial batteries, and LMT batteries above defined thresholds. AI systems that generate, update, or validate BDP data — including AI SoH estimates entered into the BDP — have a direct Art.110 compliance footprint.

What Art.110 Actually Changes

Article 110 is a targeted legislative amendment. Like Art.104–109, it inserts EU AI Act references into the conformity framework of an existing sector regulation.

The operative mechanism is the Annex I bridge. EU AI Act Annex I lists Union harmonisation legislation whose product scope intersects with the AI Act's high-risk classification pathway. Art.110 adds Directive 2006/66/EC (and by extension its successor, Regulation 2023/1542) to that list. Battery AI systems that function as safety components, health monitoring controllers, or protection systems within battery products covered by the batteries framework become Annex I battery AI — and therefore follow Art.6(1)'s high-risk classification rule when performing a safety function within the product.

Three conditions determine whether a battery AI system becomes high-risk under Art.110:

1. The host product is a battery within the scope of Directive 2006/66/EC or its successor Regulation 2023/1542
2. The AI system functions as a safety component of that battery product or is itself a battery management or monitoring system in AI-enabled form
3. The battery product with embedded AI is required to undergo third-party conformity assessment under the applicable battery regulation or a sector regulation referencing the battery

Where all three conditions are met, the developer faces dual compliance: the battery conformity process plus the full EU AI Act Title III stack.

Definitively High-Risk Battery AI Systems

AI Battery Management Systems for EV Traction Batteries

AI-driven battery management systems embedded in electric vehicle traction battery packs are definitively high-risk. The BMS performs safety-critical functions: cell voltage and temperature monitoring, overcharge and over-discharge protection, cell balancing, thermal management actuation, and state estimation. An AI BMS that incorrectly estimates cell temperature, miscalculates safe charge limits, or fails to trigger thermal protection in time creates conditions for thermal runaway — a fire or explosion risk in an occupied vehicle.

EV battery packs are simultaneously covered by Art.107 (motor vehicle safety, Regulation 2019/2144) and Art.110 (batteries). AI systems embedded in the EV traction battery sit within both Annex I bridges. The more demanding conformity assessment requirement applies. Under UNECE Regulation 100 (electric vehicle battery safety) and IEC 62660 (Li-ion cell testing), the AI BMS must demonstrate safe operation across full operational conditions.

AI Thermal Runaway Detection for Grid-Scale Storage

Stationary battery energy storage systems (BESS) used for grid frequency regulation, renewable energy integration, or industrial peak shaving are critical infrastructure. AI systems that monitor multi-megawatt-hour battery installations for early thermal runaway warning — analysing voltage signatures, gas sensor data, temperature gradients, and acoustic emissions — are definitively high-risk. A missed thermal runaway event in a grid-scale BESS can result in catastrophic fire, loss of grid stability, and risk to emergency responders and neighbouring populations.

Applicable standards: IEC 62933-5-2 (stationary battery energy storage safety), UL 9540A (fire propagation testing for BESS), NFPA 855 (stationary storage installation). AI thermal runaway detection systems that function as the primary safety layer — replacing or supplementing hardware fuses and circuit breakers — sit squarely within the Art.6(1) high-risk pathway.

AI State-of-Health Monitoring for Medical Device Batteries

Batteries powering life-sustaining medical devices — ventilators, implanted cardiac devices, infusion pumps, or AED units — require continuous state-of-health monitoring to prevent unexpected power failure. AI systems that estimate remaining battery capacity, predict end-of-life, and alert clinical staff to deteriorating battery health are safety-critical under both the EU Medical Device Regulation (MDR, Regulation 2017/745) and Art.110 of the EU AI Act. An AI SoH system that fails to alert on a degraded ventilator battery creates direct patient safety risk.

Medical device battery AI sits at the intersection of Art.110 (batteries) and Art.111(e) of the EU AI Act Annex III (safety components of medical devices). The Medical Device Regulation conformity pathway governs the primary certification, with the EU AI Act high-risk obligations layered on top.

Conditionally High-Risk Battery AI Systems

AI State-of-Health Estimation for EV Resale and Warranty

AI SoH estimation tools used for EV battery second-hand valuation, warranty adjudication, or lease residual value calculation are conditionally high-risk. The condition is whether the AI SoH estimate directly affects a safety decision — such as authorising continued vehicle operation with a degraded battery — or is used purely for commercial purposes without safety implications.

AI Second-Life Battery Assessment

The repurposing of EV traction batteries for stationary grid storage — second-life battery applications — requires accurate AI assessment of residual health, safety condition, and suitable capacity ranges. AI systems that certify a battery as safe for second-life deployment in a grid storage facility perform a safety-function determination: an incorrect clearance of an unsafe battery for grid storage installation creates fire and explosion risk in a non-automotive context with different thermal management and suppression infrastructure.

AI second-life assessment tools used by battery repurposers such as Renault-Lheritier, Volkswagen Battery Remarketing, or specialist firms (Batt'Ip, Aceleron) sit in the conditional high-risk category. The condition is whether the AI assessment output directly gates deployment into a safety-critical application.

AI Charge Optimisation for Industrial Motive Batteries

AI systems that optimise charging profiles for industrial forklift and automated guided vehicle (AGV) batteries are conditionally high-risk when the AGV or forklift operates in shared human-robot workspaces. Overcharging or thermal mismanagement caused by an AI charging optimiser that overrides manual charge limits creates occupational safety risk. The condition is whether the battery is in a product covered by sector safety legislation (Machinery Directive, AGV safety standards) linking battery safety to product safety obligations.

Provider-Deployer Split in Battery AI

Battery AI compliance responsibility is distributed across a complex supply chain involving cell manufacturers, battery pack OEMs, vehicle OEMs, analytics platform providers, and operators.

Battery OEMs as Providers: CATL, Panasonic, Samsung SDI, LG Energy Solution, and SVOLT supply integrated battery packs with embedded BMS firmware. When the BMS incorporates AI models for cell balancing, thermal management, or SoH estimation, the battery OEM is the EU AI Act provider. They must maintain technical documentation for the embedded AI system, implement a QMS under Art.17, and ensure the AI system undergoes conformity assessment before the battery pack is supplied to vehicle OEMs or industrial customers.

EV OEMs as Providers or Deployers: Tesla, Volkswagen, BMW, Stellantis, and Renault integrate battery packs into vehicles. Where the vehicle OEM develops proprietary AI BMS software layered on top of a third-party battery pack, they become the provider of the AI component. Where they integrate a battery OEM's pre-certified AI BMS without modification, they act as deployers with Art.26 obligations: due diligence on the battery OEM's conformity documentation, input data governance, and incident reporting to the battery OEM under Art.26(5).

Battery Analytics Platforms as Providers: Companies like Aviloo (EV battery testing), Recurrent (SoH ratings for used EVs), Batemo (battery simulation), and Voltu (fleet battery health) provide AI-driven SoH estimation as a service. These platforms are EU AI Act providers if their SoH output influences safety decisions, warranty determinations, or regulatory compliance (Battery Digital Passport entries).

Grid Operators as Deployers: Utility companies and grid operators deploying second-life battery storage systems (Vattenfall, Enel, Statkraft) act as deployers of AI battery monitoring systems. Their Art.26 obligations include logging and monitoring AI system outputs, ensuring only Art.110-compliant AI systems are deployed, and reporting anomalies and near-misses to system providers.

CLOUD Act Intersection for Battery AI

Battery telemetry data — cell voltage, temperature, SoC, SoH estimates, charge cycles — is increasingly processed in cloud infrastructure operated by US hyperscalers, creating CLOUD Act exposure for EU battery AI deployments.

EV Fleet Telemetry: Tesla transmits battery telemetry to AWS infrastructure. Rivian uses AWS. Volkswagen's Cariad software unit processes battery data on Microsoft Azure. Under the US CLOUD Act, US law enforcement can compel US cloud providers to produce data regardless of where it is stored. For fleet operators processing employee vehicle data or commercially sensitive battery performance data, this creates GDPR Art.46 transfer risk and competitive intelligence exposure.

SoH Model Training: Battery AI providers training SoH models on proprietary cycling data — capacity fade curves, impedance spectroscopy results, accelerated ageing test datasets — often use AWS SageMaker, Azure Machine Learning, or Google Vertex AI. The training data may include proprietary battery cell performance characteristics that are trade secrets. CLOUD Act subpoenas targeting the AI provider's cloud environment could expose competitor-sensitive training datasets.

Battery Digital Passport Data: Regulation 2023/1542 requires that Battery Digital Passport data — including AI-generated SoH values, capacity history, and material composition data — be accessible via an EU data service. If AI systems generating BDP data operate in non-EU cloud environments without adequate transfer safeguards, the BDP data pipeline has both GDPR Art.46 exposure and potential conflict with the Batteries Regulation's data governance requirements.

Sovereign deployment pathway: Battery AI providers serving regulated markets — grid operators, vehicle OEMs under fleet agreements with EU public authorities, or medical device battery applications — should route AI inference workloads through EU-jurisdiction cloud infrastructure to eliminate CLOUD Act exposure and satisfy emerging data localisation expectations under Regulation 2023/1542.

Technical Standards Intersection

Battery AI systems must be evaluated against both EU AI Act technical documentation requirements and applicable battery safety standards:

AI systems integrated into battery products must maintain technical documentation under EU AI Act Art.11 and Annex IV that explicitly cross-references the applicable battery safety standard and maps the AI system's functional scope to the standard's safety requirements.

Python Battery AI Compliance Tracker

from dataclasses import dataclass, field
from typing import Literal

BatteryCategory = Literal[
    "ev_traction", "grid_storage", "medical_device",
    "industrial_motive", "portable", "automotive_sli"
]
CloudProvider = Literal["aws", "azure", "gcp", "eu_sovereign", "on_premise"]

@dataclass
class BatteryAISystem:
    name: str
    category: BatteryCategory
    ai_function: str
    cloud_provider: CloudProvider
    triggers_safety_decision: bool
    requires_third_party_assessment: bool
    bdp_data_contributor: bool = False
    second_life_gating: bool = False

class BatteryAIComplianceTracker:
    def __init__(self):
        self.systems: list[BatteryAISystem] = [
            BatteryAISystem(
                "EV Traction BMS (CATL Qilin / Tesla BMS)",
                "ev_traction",
                "AI cell balancing, thermal management, SoH",
                "aws",
                triggers_safety_decision=True,
                requires_third_party_assessment=True,
                bdp_data_contributor=True,
            ),
            BatteryAISystem(
                "Grid BESS Thermal Runaway Detector (Fluence, BYD)",
                "grid_storage",
                "AI thermal anomaly detection, early warning",
                "azure",
                triggers_safety_decision=True,
                requires_third_party_assessment=True,
            ),
            BatteryAISystem(
                "Medical Device Battery SoH Monitor",
                "medical_device",
                "AI remaining capacity estimation, end-of-life alert",
                "eu_sovereign",
                triggers_safety_decision=True,
                requires_third_party_assessment=True,
            ),
            BatteryAISystem(
                "EV SoH Platform (Aviloo / Recurrent)",
                "ev_traction",
                "AI state-of-health estimation for resale/fleet",
                "aws",
                triggers_safety_decision=False,
                requires_third_party_assessment=False,
                bdp_data_contributor=True,
            ),
            BatteryAISystem(
                "Second-Life Battery Assessor (Renault-Lheritier)",
                "grid_storage",
                "AI safety clearance for second-life deployment",
                "azure",
                triggers_safety_decision=True,
                requires_third_party_assessment=True,
                second_life_gating=True,
            ),
            BatteryAISystem(
                "Industrial AGV Battery Charge Optimiser",
                "industrial_motive",
                "AI charge profile optimisation for AGV fleets",
                "on_premise",
                triggers_safety_decision=False,
                requires_third_party_assessment=False,
            ),
            BatteryAISystem(
                "Consumer Laptop SoC Estimator",
                "portable",
                "AI state-of-charge estimation for display",
                "on_premise",
                triggers_safety_decision=False,
                requires_third_party_assessment=False,
            ),
            BatteryAISystem(
                "SLI Battery Predictive Maintenance (OBD-linked)",
                "automotive_sli",
                "AI SoH for conventional vehicle start reliability",
                "gcp",
                triggers_safety_decision=False,
                requires_third_party_assessment=False,
            ),
        ]

    def classify(self, s: BatteryAISystem) -> str:
        if s.category in ("ev_traction", "medical_device") and s.triggers_safety_decision:
            return "DEFINITELY_HIGH_RISK"
        if s.category == "grid_storage" and s.triggers_safety_decision:
            return "DEFINITELY_HIGH_RISK"
        if s.second_life_gating and s.triggers_safety_decision:
            return "DEFINITELY_HIGH_RISK"
        if s.triggers_safety_decision or s.requires_third_party_assessment:
            return "CONDITIONALLY_HIGH_RISK"
        return "NOT_HIGH_RISK"

    def cloud_act_exposure(self, s: BatteryAISystem) -> bool:
        return s.cloud_provider in ("aws", "azure", "gcp")

    def compliance_summary(self) -> dict:
        definitely = [s for s in self.systems if self.classify(s) == "DEFINITELY_HIGH_RISK"]
        conditional = [s for s in self.systems if self.classify(s) == "CONDITIONALLY_HIGH_RISK"]
        not_high = [s for s in self.systems if self.classify(s) == "NOT_HIGH_RISK"]
        cloud_risk = [s for s in self.systems if self.cloud_act_exposure(s)]
        bdp = [s for s in self.systems if s.bdp_data_contributor]
        return {
            "total": len(self.systems),
            "definitely_high_risk": len(definitely),
            "conditionally_high_risk": len(conditional),
            "not_high_risk": len(not_high),
            "cloud_act_exposure": len(cloud_risk),
            "bdp_contributors_needing_review": len(bdp),
            "dual_compliance_required": len(definitely) + len(conditional),
        }

tracker = BatteryAIComplianceTracker()
summary = tracker.compliance_summary()
# {"total": 8, "definitely_high_risk": 3, "conditionally_high_risk": 2,
#  "not_high_risk": 3, "cloud_act_exposure": 5, "bdp_contributors_needing_review": 2,
#  "dual_compliance_required": 5}

The Battery Digital Passport and AI Compliance

Regulation (EU) 2023/1542 introduces the Battery Digital Passport (BDP) — a mandatory digital record for EV batteries, industrial batteries above 2 kWh, and light means of transport (LMT) batteries above defined thresholds. The BDP must include battery state-of-health data, capacity history, cycle count, and material composition. For AI systems contributing data to the BDP, this creates a specific EU AI Act compliance footprint:

AI SoH estimates in the BDP: If an AI system generates the SoH value that is recorded in a Battery Digital Passport and subsequently used to gate second-life deployment or assess residual value, the AI output directly determines a regulatory compliance decision. This brings the SoH AI system within the Art.6(1) high-risk pathway via the Art.110 bridge.

BDP data integrity obligations: Art.15 (accuracy and robustness under Art.15(a)) and Art.11 (technical documentation) require that AI systems feeding BDP data maintain traceable, auditable output histories. BDP data cannot be retroactively corrected without audit trail implications — AI SoH systems contributing to BDPs must implement versioned output logging aligned with Art.72 post-market monitoring requirements.

BDP cross-border data access: Regulation 2023/1542 requires BDP data to be accessible to EU authorities, recyclers, and repurposers via a registered data service. AI providers operating BDP-contributing SoH platforms must ensure their data service architecture is accessible from EU jurisdictions and not exclusively hosted behind CLOUD Act-exposed US infrastructure.

The Art.104–112 Amendment Series

Article 110 is the sixth in the Art.104–112 series of EU AI Act sector amendments, each inserting the high-risk pathway into existing EU product regulation frameworks:

The Art.110 batteries amendment introduces a compliance complexity unique in the series: the underlying directive is actively being superseded by a new regulation (2023/1542) during the same timeframe as the EU AI Act's application date. Battery AI developers must track both the EU AI Act compliance timeline (August 2026 for high-risk systems) and the Batteries Regulation application milestones to ensure they remain within the correct conformity framework. No other article in the Art.104–112 series faces this dual-transition challenge.

25-Item Battery AI Compliance Checklist (Art.110)

Product Scoping

1. Confirm the battery falls within Directive 2006/66/EC or Regulation 2023/1542 scope (portable, automotive, industrial, EV traction)
2. Identify all AI systems embedded in or controlling the battery (BMS AI, SoH estimator, thermal monitor)
3. Determine whether the battery is subject to third-party conformity assessment under applicable sector regulation (UNECE Reg.100 for EVs, MDR for medical devices)
4. Map AI systems to safety functions: thermal protection, over-charge prevention, end-of-life alerting, second-life gating
5. Identify Battery Digital Passport obligations under Regulation 2023/1542 and whether any AI system generates BDP-submitted data

High-Risk Classification 6. Apply Art.6(1) test: is the AI a safety component of a batteries-directive-scoped product requiring notified body assessment? 7. Confirm Art.110 Annex I bridge applies to your specific battery category 8. Check for overlapping sector regulation coverage (Art.107 for EV traction batteries also covered by Reg.2019/2144) 9. Assess whether AI SoH estimates feed regulatory decisions (BDP, warranty, second-life clearance) triggering high-risk classification 10. Document the classification decision with regulatory basis covering both the Batteries Directive/Regulation and the EU AI Act

Dual Conformity Assessment 11. Plan batteries conformity path and identify applicable battery safety standard (IEC 62619, ISO 6469-1, UNECE Reg.100) 12. Identify notified body competent for battery products and EU AI Act assessment 13. Establish EU AI Act technical documentation (Art.11, Annex IV) for each high-risk battery AI system 14. Implement AI risk management system (Art.9) with battery-specific failure mode analysis (thermal runaway, SoH underestimation, false end-of-life) 15. Define human oversight mechanisms (Art.14) for battery safety decisions and SoH-gated operational authorisations

Battery Digital Passport 16. Map all AI-generated data flows to BDP fields: identify which AI outputs become regulatory BDP records 17. Implement versioned AI output logging aligned with Art.72 post-market monitoring for BDP-contributing systems 18. Ensure BDP data service architecture is accessible from EU jurisdictions without CLOUD Act-exposed intermediaries 19. Define AI model update governance for BDP contributors: retrain triggers, output drift thresholds, BDP correction procedure 20. Assess GDPR Art.9 implications for BDP data combining battery location, identity, and health history

Data and Cloud 21. Inventory battery telemetry data flows to cloud platforms: identify CLOUD Act exposure for EV and BESS data 22. Assess SoH model training data sovereignty: is training data processed exclusively in EU-jurisdiction infrastructure? 23. Evaluate data localisation requirements for grid storage AI in regulated energy markets 24. Map medical device battery SoH data against MDR data requirements and EU AI Act Art.10 data governance 25. Register high-risk battery AI systems in EU AI Act database (Art.71) before August 2026 application deadline

See Also

• EU AI Act Art.111: Amendment to Directive 2014/68/EU — AI Structural Integrity Monitoring, NDT Analysis and Pressure Vessel Inspection High-Risk Classification
• EU AI Act Art.109: Amendment to Directive 2014/53/EU — Software-Defined Radio, Cognitive Radio and AI Spectrum Management High-Risk Classification
• EU AI Act Art.108: Amendment to Directive 2014/90/EU — Marine Equipment ECDIS and Autopilot AI
• EU AI Act Art.107: Amendment to Regulation (EU) 2019/2144 — ALKS, ADAS and Motor Vehicle AI High-Risk Classification
• EU AI Act Art.106: Amendment to Regulation (EU) 168/2013 — L-Category Vehicles Motorcycle ABS and Rider AI
• EU AI Act Art.105: Amendment to Regulation (EU) 167/2013 — Agricultural Vehicles Autonomous Guidance AI