Legal issues in automated vehicles: critically considering the potential role of consent and interactive digital interfaces

The paper examines partially automated vehicles that share operational responsibility with human drivers and the legal need to support safe handovers and predictable liability after accidents. A proposed mechanism is to present operating conditions and liability terms via an in-vehicle interactive digital interface and obtain the driver’s consent before activation. The authors argue that human factors in automation and the way people interact with digital interfaces undermine the elements required for valid consent, jeopardizing attempts to use consent to allocate liability. The introduction highlights empirical evidence of risky automated-to-manual transitions, technological hype around AV safety, and the potential for drivers to misjudge risk and over-trust automation. The paper’s purpose is to provoke regulatory and industry discussion on whether driver consent collected through digital interfaces can meaningfully and lawfully delineate responsibility and to explore alternatives, including specialized training (e.g., PASCAL/Guide2Autonomy) to calibrate drivers’ understanding of risks and responsibilities.

Human factors research indicates significant delays in regaining situational awareness during AV-to-human handovers, ranging from at least 8 seconds to up to 40 seconds (Agrawal et al., 2017; Merat et al., 2014). Drivers may be poor monitors when engaged in secondary tasks and can develop over-trust in reliable automation, leading to complacency and misuse (Lee & See, 2004; Singh et al., 1997; Alonso Raposo et al., 2017). Reaction times and readiness vary with individual characteristics and conditions (Fofanova & Vollrath, 2011), and drivers often misjudge their readiness to take over (Saito et al., 2018). Technological mitigations include safe-stop maneuvers and driver monitoring via physiological and behavioral sensors (Magdici & Althoff, 2016; Alrefaie et al., 2019; Svensson, 2018), yet alerts may fail to restore attention effectively and can have adverse effects (van den Beukel et al., 2016), and sensor interpretation and calibration remain challenging (Collett & Musicant, 2019). The paper synthesizes literature on digital consent and user interfaces, noting that people commonly ignore electronic terms and conditions and safety information (Solove, 2012/13; Obar & Oeldorf-Hirsch, 2020), that persuasive design and nudging can steer behavior (Fogg, 2003; Schneider et al., 2019), and that even safety videos (e.g., in aviation) suffer from low attention and retention (Ragan et al., 2017). Legal sources reviewed include the Vienna Convention amendment enabling automated systems (UNECE, 2016), the UK Automated and Electric Vehicles Act 2018 emphasizing driver responsibilities, and evolving UNECE guidelines placing obligations on drivers to be capable of takeover and aware of rules (Global Forum for Road Safety, 2020). Alternative liability conceptualizations recognize concurrent, interleaved responsibilities of driver and AV (Bellet et al., 2019), but assume robust detection of driver readiness and reliable safe-stop—assumptions the literature questions.

The study is a theoretical and conceptual analysis integrating legal doctrine on consent and liability with empirical findings from human factors and HCI/UX research. It critically evaluates the feasibility of using driver consent, communicated via an interactive digital interface, to allocate operational responsibility and legal liability in partially automated vehicles. The authors draw on a narrative review of psychological and ergonomics literature on takeover performance, trust, attention, and interface design, as well as statutory and regulatory sources (Vienna Convention amendments, UK AEVA 2018, UNECE guidelines). The paper also considers ongoing applied research (the PASCAL project and its Guide2Autonomy framework) as a prospective component of driver education and certification, but it does not conduct new empirical experiments.

1. Human factors evidence shows drivers commonly experience substantial delays in regaining situational awareness during automated-to-manual transitions (minimum ≈8 s; up to ≈40 s), making handovers hazardous. Placing hands on the wheel can occur quickly (≈1.5 s) but does not equate to cognitive readiness. 2) Drivers often misperceive their readiness and may over-trust reliable automation, especially when multi-tasking; monitoring performance degrades with secondary tasks. 3) Driver monitoring and safe-stop technologies can mitigate risk but are imperfect; alerts may not restore attention effectively, and safe stops are not always feasible. 4) Users frequently ignore or skim complex information, warnings, and legal terms delivered via digital interfaces; persuasive, seamless UI design may further discourage deep processing, undermining informed consent. 5) Given these factors, consent obtained through an in-vehicle digital interface is unlikely to meet the legal standards of voluntariness and informed understanding necessary to pre-allocate liability reliably; such consent may not withstand judicial scrutiny after accidents. 6) Current and draft legal frameworks place heavy responsibility on drivers (e.g., UK AEVA 2018; UNECE guidance), exacerbating asymmetry if consent mechanisms disadvantage drivers. 7) Specialized AV driver training and potentially certification can improve mental models, calibrated trust, and takeover performance (simulators and VR can help but real-vehicle training is likely necessary). 8) Liability and insurance predictability will remain weak unless driver comprehension of risk and responsibility is demonstrably improved and interface communications are validated for effectiveness.

The analysis addresses whether driver consent via digital interfaces can legitimately and predictably transfer operational responsibility and liability between drivers and partially automated vehicles. Empirical human factors data suggest that drivers face inherent cognitive and temporal limitations during handovers, and HCI research indicates that digital interfaces tend to elicit shallow engagement with complex terms, undermining informed consent. Consequently, relying on interface-mediated consent to assign liability is both unsafe and legally fragile. The findings imply that regulators and manufacturers should not presume that a driver’s click-through agreement establishes a fair or enforceable delineation of responsibility. Instead, a fair culpability framework requires: (a) robust, human-centered system design including reliable detection of driver readiness and safe fallback behaviors; (b) validated communication strategies that demonstrably convey risks and obligations; and (c) specialized driver education to develop calibrated trust and takeover competence. The Guide2Autonomy concept exemplifies a path toward user-centric training and evaluation, integrating simulation, real-world trials, and feedback loops. Aligning training, certification, and interface design with legal standards of informed consent can enhance safety and improve the predictability of liability and insurance outcomes.

The paper concludes that using driver consent obtained via an interactive digital interface as a mechanism to transfer operational responsibility and pre-allocate legal liability in partially automated vehicles is unlikely to be effective or enforceable. Fundamental human factors constraints during handovers and well-documented weaknesses in how users process electronic warnings and legal terms make informed consent doubtful in this context. Manufacturers cannot absolve themselves of liability merely through interface-mediated agreements. To create a fair and predictable liability framework, the authors recommend specialized AV driver training and possible certification to build accurate mental models, calibrated trust, and takeover skills; and further research to validate the substance and efficacy of interface communications regarding risks, responsibilities, and legal conditions. The PASCAL Guide2Autonomy framework is presented as a way to advance user-centric training and evaluation that could support more robust legal and insurance frameworks.

The work is a theoretical, conceptual examination without new empirical testing of drivers interacting with AV interfaces. The translation of known user behavior with online terms and airline safety briefings to in-vehicle AV interfaces is inferred rather than directly measured; empirical studies are needed to validate how AV drivers comprehend and act on interface-delivered warnings and legal conditions. The technological state of AV driver monitoring and safe-stop capabilities is evolving, and conclusions may change as systems mature. Differences across manufacturers and interface designs may limit generalizability, and the proposed training and certification concepts (e.g., Guide2Autonomy) are under development and not yet validated at scale.