Perceived opportunities and risks of using deepfake technology in grief treatment: a qualitative study among Dutch bereaved people and mental health care workers

ABSTRACT

Background: Deepfake technology, capable of simulating realistic representations of deceased people, is receiving growing attention as a potential technological tool in grief treatment. While its use has been explored in trauma treatment, its application in grief treatment remains unstudied.

Objective: We examined how Dutch bereaved people and mental healthcare workers (MHCWs) perceive the usefulness and risks of using deepfake technology in grief treatment.

Method: After viewing a videoclip of a deepfake grief intervention, participants listed three reasons for and against its use. We conducted a qualitative content analysis of open-ended survey responses from 159 bereaved people (86% women, Mage = 55.09) and 126 MHCWs (81% women, Mage = 39.75).

Results: Bereaved people found deepfakes emotionally meaningful, envisioning them as a way to say goodbye, reconnect, or find closure. MHCWs acknowledged therapeutic potential, particularly as a supplement to established techniques (e.g. exposure, imagery rescripting), but highlighted ethical issues regarding misrepresentation, informed consent, and client safety. Both groups raised concerns about psychological risks, including hindrance of loss acceptance and emotional dependency, especially for vulnerable clients.

Conclusions: Bereaved people valued opportunities for reconnection with the deceased, while MHCWs warned that such reconnection might impede grief processing. Both groups agreed that deepfakes may have therapeutic value if applied ethically, under therapist guidance, and with safeguards. Because participants were asked about hypothetical use, results reflect perceptions, not clinical effectiveness. Our findings underscore the need for ethical guidelines to balance bereaved people’s needs with professional responsibility when exploring the therapeutic effects of deepfake technology in treatment.

HIGHLIGHTS

• 749 textual responses from bereaved people and 654 textual responses from mental health care workers regarding motivations to use or not use deepfake technology in grief treatment were qualitatively analysed.

• One in four bereaved people and one in three mental health care workers expressed willingness to use deepfake technology in grief treatment.

• Both groups viewed deepfake technology as potentially valuable for addressing ‘unfinished business’.

1. Introduction

Prolonged grief disorder (PGD) includes intense longing for, and/or preoccupation with, the deceased and other symptoms disrupting daily functioning, at least six (in ICD-11; World Health Organization [WHO], 2019) or 12 months after the death of a loved one (in DSM-5-TR; American Psychiatric Association [APA], 2022). Population-based studies indicate that 3–5% of bereaved people meet the criteria for self-rated ICD-11 and DSM-5-TR PGD (Rosner et al., 2021; Treml et al., 2024). There is empirical evidence that PGD can be treated. However, there is insufficient evidence that PGD can be prevented (Avis et al., 2025; Wittouck et al., 2011).

Cognitive–behavioural therapy (CBT) and Complicated Grief Therapy (CGT) are examples of evidence-based treatments for PGD, which typically focus on helping people accept the reality of loss, reduce avoidance behaviour, and engage in meaningful activities (Boelen et al., 2007; Bryant et al., 2014; Lenferink et al., 2023; Reitsma et al., 2023; Rosner et al., 2014; Shear et al., 2005; Shear et al., 2014; Shear et al., 2016; Szuhany et al., 2021). However, the effectiveness of treatments remains small to moderate (Johannsen et al., 2019; Komischke-Konnerup et al., 2021), and a significant proportion of clients do not recover fully. Overall, this suggests room for improvement in grief treatment, sparking the possibility of introducing new technologies, such as deepfake technology, to enhance existing evidence-based techniques.

Deepfake technology utilises artificial intelligence, including machine learning and deep learning, to generate and manipulate audio and visuals that appear authentic but are fake. If applied in grief treatment, deepfake technology could be used to manipulate a picture of the deceased, utilising facial mapping technology, where the deepfake image of the deceased person's face is controlled by the therapist's voice (Westerlund, 2019). This technique allows the deepfake representation of the deceased to appear as though they are speaking (see Figure 1).

Figure 1.

Process of deepfake-assisted therapy.

Notes: This figure outlines the phases of preparation, creation, operation, and application of deepfake-assisted therapy. Panel A: The client (left) provides a photograph of the deceased to the therapist (right). Panel B: The therapist uploads this photograph to create the deepfake. Panel C: The therapist controls the deepfake with their voice. The voice causes the deepfake animation to make mouth movements that mimic the voice. Panel D: During therapy, the client (left) and therapist (right) sit in different rooms (indicated by the dotted line), each in front of a monitor. The client engages with the deepfake deceased, controlled by the therapist, displayed on their screen. Figure generated using Sora image generation AI.

Deepfake therapy has yielded promising results in trauma treatment in a study by van Minnen et al. (2022). They explored and applied the use of deepfake technology as a tool for perpetrator confrontation for sexual violence-related posttraumatic stress disorder (PTSD) and moral injury. They found that these realistic simulations may help victims confront and access complex emotions that were otherwise difficult to access using traditional methods.

Using deepfake may also be helpful in grief treatment as a technology supporting existing evidence-based therapeutic techniques. Deepfake technology enables a realistic way of exposing clients to painful emotions, memories, or images related to the loss, which could be particularly useful for clients struggling with traditional techniques in grief and trauma treatment, like the empty-chair method, imaginary exposure, imagery rescripting, and letter-writing (Bryant et al., 2014; Field & Filanosky, 2009; Hagl et al., 2015; Lechner-Meichsner et al., 2024; Shear, 2015; Wagner et al., 2005).

Across these existing evidence-based techniques, two shared therapeutic ingredients are used, namely structured exposure to loss-related stimuli (e.g. emotions, memories, or images) and therapist-guided interpersonal processing (e.g. imaginal dialogue), which may help clients articulate and integrate distressing loss-related experiences (Boelen et al., 2006; Maccallum & Bryant, 2013; Shear, 2010). Furthermore, bereaved people with elevated levels of unfinished business (i.e. unexpressed or unresolved issues between the bereaved person and the deceased; Holland et al., 2018) or unsatisfactory goodbyes, which are relatively common after sudden or violent loss (Klingspon et al., 2015), may use deepfake therapy to have virtual interactions with the deepfaked deceased, potentially finding a sense of emotional resolution that was not possible in real life. In this sense, deepfake technology can be conceptualised as a therapist-guided technological tool, and not a stand-alone treatment that may make imaginal dialogue and imagery-based techniques related to communicating with the deceased more concrete and emotionally evocative by providing realistic visual and auditory cues.

Despite these promising possibilities, there are ethical concerns regarding the use of deepfake technology in grief treatment. First, it may blur the line between reality and fiction when simulating a deceased loved one. This could result in difficulties in distinguishing between the fake interaction and people's memories. Second, the use of deepfake technology in grief treatment may fuel maladaptive grieving patterns (Hoek et al., 2025). More specifically, a simulated conversation with the deepfake of a deceased may offer initial comfort, but could eventually foster dependency and interfere with acceptance of the loss. This aligns with grief theories stating that approach or proximity-seeking behaviours towards the deceased could serve as avoidance of the reality of the irreversibility of the loss (Boelen et al., 2006; Eisma & Lenferink, 2023; Shear, 2012). These insights underscore the importance of carefully considering both the drawbacks and benefits of utilising deepfake technology in grief treatment.

To examine the potential usefulness of deepfake technology in grief treatment, it is essential to gather perspectives from both bereaved people and Mental Health Care Workers (MHCWs). Earlier research highlighted the importance of including both end-users when developing and implementing new tech-supported mental health interventions to ensure safety, effectiveness, and relevance (Egan et al., 2024; Lenferink et al., 2021; Mohr et al., 2017; Torous et al., 2018; van Dijk et al., 2025).

Understanding stakeholder perspectives is essential given the ethical and clinical challenges associated with introducing deepfake technology into bereavement care. This study therefore aimed to map perceived reasons for and against the use of deepfake technology in grief treatment, as reported by bereaved adults and MHCWs. By identifying perceived opportunities and risks across these stakeholder groups, the findings are intended to inform future research.

2. Methods

2.1. Participants

This cross-sectional survey study is part of a larger project focusing on perceptions of Dutch bereaved people and MHCWs on the usefulness and nonsense of using deepfake technology in grief treatment. All participants gave informed consent. The study received approval from the Ethics Committee of the University of Twente (ID: 240204). Dutch people were eligible to participate if their partner, family member, or friend died at least 12 months earlier. No inclusion criteria were formulated for PGD intensity in order to include a heterogeneous sample.

For MHCWs, participation was open to people who were actively treating clients in the field of mental health. We did not specify inclusion criteria regarding specific areas of expertise and years of experience. However, MHCWs were required to spend at least one hour a week actively treating clients.

For bereaved people, invitation emails were sent to those who had completed a self-test to monitor their grief reactions on a Dutch website providing information about grief (treatment) (www.rouwbehandeling.nl). From these, a random selection of eligible people who agreed to be contacted for future research was invited to participate. Additionally, bereaved people were encouraged to share the invitation with others.

MHCWs were contacted via email through the researchers’ networks. Several Dutch professional organisations for psychologists shared the invitation with their members. Flyers with study information were posted on social media platforms, such as LinkedIn. MHCWs were additionally asked to share the invitation with their colleagues. This invitation was titled ‘The sense and non-sense of using deepfake technology in grief treatment’, with the goal to recruit bereaved people and MHCWs for and against the idea of using deepfake technology in grief treatment to obtain both perspectives.

2.1.1. Procedures

Both bereaved people and MHCWs completed online surveys between April and June 2024. Once participants used the link, they were directed to the survey, where they were asked to provide information about their background, the loved one who had died (for bereaved people), and their professional background (for MHCWs) .¹ Participants were instructed to view a short 27 s videoclip from a Dutch documentary (Nejal, 2020). This videoclip depicts bereaved parents engaging in a therapist-guided, live deepfake-interaction with their deceased daughter (see Appendix 1 for the instruction and transcript of the videoclip). This videoclip was used as a standardised primer to provide a shared reference across participants for what such an interaction could look like in a therapeutic context. In this documentary, state-of-the-art deepfake technology is used (https://deeptherapy.ai/), which is the same as was used in PTSD treatment from van Minnen et al. (2022).

2.2. Measures used for bereaved people

2.2.1. Motivations for (not) using deepfake technology

After watching the previously described videoclip, bereaved people were asked to provide their three most important reasons to use deepfake technology in grief treatment with the question: ‘What are, according to you, the three most important reasons to use deepfake technology in grief treatment?’ followed by three text boxes in which participants were able to provide their reasons. This was followed by an inquiry about their three most important reasons to not use deepfake technology in grief treatment with the question: ‘What are, according to you, the three most important reasons to not use deepfake technology in grief treatment?’²

2.2.2. Prolonged grief intensity

Bereaved people completed the 22-item Traumatic Grief Inventory – Self Report Plus (TGI-SR+), which assesses PGD intensity (Lenferink et al., 2022). They rated the frequency of their symptoms in the past month using a five-point scale (1 = never, 2 = rarely, 3 = sometimes, 4 = frequently, 5 = always). Total scores were created by summing all items. A score of ≥71 was considered a probable PGD case. The Dutch TGI-SR + has good psychometric properties (Lenferink et al., 2022). Cronbach’s alpha in this study was α = .94.

2.2.3. Sociodemographic and loss-related characteristics

Questions related to background characteristics were: gender (1 = man, 2 = woman, 3 = other), age, country of birth, and highest obtained educational level (1 = primary school, 2 = high school, 3 = vocational education, 4 = college or university). Bereaved people indicated whether they had received support from a psychologist, therapist, or psychiatrist before the death of their loved ones (1 = yes, 2 = no), had received support related to the death of their loved ones, and whether they were currently receiving support related to the death of their loved ones (1 = yes, 2 = no). Additionally, we inquired if the participant had previously used deepfake technology in grief treatment (1 = yes, 2 = no). Bereaved people also responded to the following statement: ‘If I had access to deepfake technology in grief therapy, I would use it.’ with answer options 1 = completely disagree, 2 = disagree, 3 = neutral, 4 = agree, and 5 = completely agree. The latter two answer options were used as indicators for willingness to use deepfake technology in grief treatment. Considering the loss-related characteristics, we asked about the relationship to the deceased, the age of the deceased at the time of death, the date of death, cause of death (1 = physical illness, 2 = accident, 3 = suicide, 4 = homicide or manslaughter, 5 = other, namely), and perceived unexpectedness of the death (1 = not unexpected, 2 = a little unexpected, 3 = unexpected, 4 = very unexpected, 5 = completely unexpected).

2.3. Measures used for MHCWs

2.3.1. Motivations for (not) using deepfake technology

The same videoclip, followed by open-ended questions, was presented to MHCWs.

2.3.2. Sociodemographic and professional characteristics

Background characteristics were gender (1 = man, 2 = woman, 3 = other), age, country of birth, and highest achieved level of education (1 = primary school, 2 = high school, 3 = vocational education, 4 = college or university). MHCWs were asked about their current job function, workplace, years of experience in mental healthcare since completing their studies, the average number of clients treated per week, the average number of treatment hours per week, and what psychological disorders they treated most frequently. Additionally, MHCWs were asked if they had experience with treating grief symptoms (1 = yes, 2 = no) and, if yes, were asked about their average time treating grief symptoms per week in percentages. Lastly, MHCWs were asked if they had ever used deepfake technology in psychological treatment (1 = yes, 2 = no) and, if so, whether they had used it in grief treatment (1 = yes, 2 = no). MHCWs also responded to the following statement: ‘If I had access to deepfake technology in grief therapy, I would use it.’ using the same 5-point Likert scale as for bereaved people. Again, the highest two answer options were used as indicator for willingness to use deepfake technology in grief treatment.

2.4. Data analysis

After cleaning the data (for details see Appendix 2), we conducted an inductive qualitative content analysis of open-ended responses (Graneheim & Lundman, 2004; Hsieh & Shannon, 2005). Responses were divided into meaningful units. Each unit was coded with a subtheme. Subthemes were subsequently grouped into overarching main themes. Subthemes were labelled as closely as possible to participants’ wording to reflect explicit content rather than impose predefined theoretical constructs. Each motivation for or against using deepfake technology in grief treatment could contain multiple units and thus multiple subthemes. For example, one participant wrote: ‘I think it helps with exposure and could help to accept loss’. This was divided into two units labelled with the subthemes ‘Exposure’ and ‘Helps accepting loss’. See our structured pseudonymised datafile linking each unit to its assigned subtheme and main theme via https://doi.org/10.17026/SS/OHG6LG.

To avoid inflation due to verbosity, each subtheme could be counted only once per participant within each prompt category (reasons for vs. reasons against). Subtheme counts therefore represent the number of participants who mentioned that subtheme at least once within a category. Main theme counts represent the sum of participant-mentions across the subthemes belonging to that main theme (i.e. a participant may contribute to multiple subthemes within a main theme). A three-stage process was used to code the open-text responses (see Appendix 3 for details and Appendix 4 for an example).

A second independent rater (the second author) independently coded a random subset of 20% of the data. Agreement was defined as the alignment in underlying meaning, regardless of exact phrasing. Any discrepancies were discussed and resolved between the first and last authors and used in subsequent discussions to refine the coding framework applied to the full dataset.

Agreement was calculated as the percentage of reasons where both raters identified overlapping subthemes, excluding blank responses. Agreement percentages are provided descriptively to indicate convergence on meaning rather than as statistical reliability estimates. For bereaved people, the reasons in favour and against using deepfake technology reached an agreement of 77% and 70%, respectively. For MHCWs, the reasons in favour and against using deepfake technology reached an agreement of 67% and 80%, respectively.

We approached the analysis from an interpretive qualitative stance, treating themes as co-constructed through engagement with participants’ accounts. Given the ethical and emotional sensitivity of the topic, we acknowledge that researchers’ backgrounds may shape interpretive decisions. The research team consisted of a senior grief researcher with extensive qualitative research experience (last author) and two psychology graduate students (first and second authors). These positions, alongside our personal experiences with loss and mental health care, informed our interpretation of the data. We adopted a critical-realist approach, treating participants’ accounts as both expressions of lived experience and as processes of meaning-making situated within broader social, cultural, and technological contexts that shape adaptation to loss. To enhance transparency and credibility, the first author conducted iterative coding under the supervision of the last author, with regular discussions to resolve ambiguities and refine coding definitions.

3. Results

3.1. Sociodemographic and loss-related characteristics of bereaved people

The 159 bereaved people were aged on average 55.09 years (SD = 12.93) and most identified as women (86%) (see Table 1). Regarding the relationship to the deceased, 64% had lost a partner and 17% a parent. The primary cause of death was physical illness (67%) followed by suicide (11%). The mean PGD score was 54.28 (SD = 16.18) and 15% scored above the cut-off for probable PGD cases. None of the bereaved people had received grief support using deepfake technology. In total, 39 bereaved people (25%) were willing to use deepfake technology in grief treatment if they had access to it.

Table 1.

Sociodemographic and loss-related characteristics of bereaved people (N = 159).

Age of participant, M (SD)	55.09	12.93
Gender, N (%)  Woman  Man  Other	137 22 0	86.20 13.80 0.00
Country of birth, N (%)  The Netherlands  Other	152 7	95.80 4.20
Education, N (%)  Elementary and secondary school  Vocational education  (Applied) University	21 39 99	13.20 24.50 62.30
Deceased is my, … . N (%)  Partner  Parent  Sibling  Child  Other	102 27 12 10 8	64.20 17.00 7.50 6.30 5.10
Cause of death, N (%)  Physical illness  Suicide  Accident  Murder or manslaughter  Other	107 17 9 1 25	67.30 10.70 5.70 0.60 15.70
Time since loss (in months), M (SD)	61.92	75.143
Age of the deceased (in years), M (SD)a	56.26	18.17
Expectancy of death, N (%)  Completely expected  A little unexpected  Quite unexpected  Very unexpected  Completely unexpected	30 28 24 20 57	18.90 17.60 15.10 12.60 35.80
Previous support from a psychologist, psychotherapist or psychiatrist preceding the loss, N (%)  No  Yes	90 69	56.60 43.40
Previous grief support from a psychologist, psychotherapist or psychiatrist, N (%)  No  Yes	58 101	36.50 63.50
Current grief support from a psychologist, psychotherapist or psychiatrist, N (%)  No  Yes	137 22	86.16 13.84
Wish to receive support from a psychologist, psychotherapist or psychiatrist, N (%)  No  Yes	132 27	83.02 16.98
Prior knowledge about deepfake technology, N (%)  No  Yes	78 81	49.10 50.90
Recruitment, N (%)  Invitation email  Social media  Referral of acquaintance  Other	133 13 4 9	83.60 8.20 2.50 5.70

Note: M = mean, SD = standard deviation.

^aOne invalid response was considered missing for one participant.

3.2. Sociodemographic and professional characteristics of MHCWs

The average age of the 126 MHCWs was 39.75 years (SD = 12.60), with 81% identifying as women (see Table 2). Nearly all participants, 125 (99%), completed (applied) university. For details regarding work experience see Appendix 5. In total, 41 MHCWs (34%) were willing to use deepfake technology in grief treatment if they had access to it.

Table 2.

Characteristics of mental health care workers (MHCWs) (N = 126).

Age, M (SD)	39.75	12.60
Gender, N (%)  Woman  Man  Other	102 22 2	81.00 17.50 1.60
Country of birth, N (%)  The Netherlands  Other	117 9	92.86 7.14
Education, N (%)  Elementary and secondary school  Vocational education  (Applied) University	0 1 125	0.00 0.80 99.20
Recruitment, N (%)  Invitation email  Social media  Referral of an acquaintance  Other	72 30 14 10	57.10 23.80 11.10 7.90
Experience with treating prolonged grief, N (%)  No  Yes	37 89	29.40 70.60
Relative time spent on treating (prolonged) grief weekly (in %), M (SD)	19.07	21.89
Experience with using deepfake technology in treatment, N (%)  No  Yes	124 2	98.40 1.60
Experience with using deepfake technology in grief treatment, N (%)  No  Yes	124 2	98.40 1.60
Prior knowledge about deepfake technology, N (%)  No  Yes	33 93	26.20 73.80

Note: M = mean, SD = standard deviation.

3.3. Frequency of reasons for and against using deepfake technology in grief treatment

Not all participants could provide three reasons in favour or against using deepfake technology in grief treatment. Table 3 shows the number of bereaved people and MHCWs who could provide 3, 2, 1 or 0 reasons.

Table 3.

Number of reasons provided by bereaved people (N = 159) and MHCWs (N = 126) for and against the use of deepfake technology in grief treatment.

N of Reasons	Reasons for using deepfake technology in grief treatment, N (%)	Reasons for not using deepfake technology in grief treatment, N (%)
Bereaved people
3	85 (53)	113 (71)
2	31 (20)	28 (18)
1	22 (14)	15 (9)
0	21 (13)	3 (2)
MHCWs
3	89 (71)	94 (75)
2	26 (21)	18 (14)
1	7 (5)	10 (8)
0	4 (3)	4 (3)

Note: MHCWs = Mental healthcare workers.

3.4. Reasons for using deepfake technology in grief treatment by bereaved people

Table 4 displays the top three main themes and the most frequently mentioned subthemes within each main theme mentioned by bereaved people as reasons for using deepfake technology in grief treatment. In some cases, less than three subthemes were included due to equal frequency of occurrence.

Table 4.

Main themes and subthemes in favour vs against using deepfake technology in grief treatment as mentioned by bereaved people (N = 159).

Reasons in favour of using deepfake technology in grief treatment
Main themes (number of occurrences)	Subthemes (number of participants referring to subtheme)
The use of deepfake can provide the opportunity to address unfinished business. (92)	Deepfake can be used to express things left unsaid to the deceased. (43) Deepfake can be used to ask unanswered questions. (25) Deepfake can be used to get answers to pressing questions. (10)
The use of deepfake can give me the chance to get in touch with the deceased. (90)	Deepfake gives the opportunity to see the deceased again in moving images. (19) Deepfake can be used to reconnect with the deceased. (18) Deepfake gives the opportunity to hear the voice of the deceased again and talk to them. (18)
The use of deepfake can help in the grieving process. (60)	The use of deepfake aids the grieving process. (28) The use of deepfake can help those with prolonged grief. (9) The use of deepfake can help accept the death of their loved one. (8)a The use of deepfake can help to move on with life. (8)a
Reasons against using deepfake technology in grief treatment
Main themes (number of occurrences)	Subthemes (number of participants referring to subtheme)
The use of deepfake is fake and artificial. (125)	The use of deepfake is fake. (43) The deepfake's answers are not truly from the deceased. (27) The deepfake is not really the deceased. (13)
The use of deepfake hinders the grieving process. (101)	The use of deepfake is too confronting. (31) The use of deepfake makes accepting the death of the deceased more complicated. (14) The use of deepfake is not needed in a normal grieving process. (10)a The use of deepfake makes it harder to let go of the deceased. (10)a With the use of deepfake, a new farewell must be taken. (10)a
The use of deepfake entails psychological risks. (82)	The use of deepfake can be addictive, as it creates a desire to keep talking to the deceased. (18) The use of deepfake may cause confusion with what is real and what is not. (13) The use of deepfake can intensify grief symptoms. (12)

^aNot mentioned in text due to equal frequency of occurrence.

The use of deepfake can provide the opportunity to address unfinished business. This main theme showed how bereaved people considered that the deepfake could help them say the things they always wanted to say to the deceased or ask the questions they never got the opportunity to ask. This main theme encompassed 92 quotations with six different subthemes.

Deepfake can be used to express things left unsaid to the deceased. This subtheme describes how participants felt they could still convey the final things they wanted to say to the deceased to resolve issues they had never had the chance to address while the deceased was alive. Participants expressed this while being aware that the deepfake was not the deceased.

If you feel that not everything was said during that person's life, you can still do so. In this way, even though you know it's not real. (Woman, aged 60–69, lost a sibling due to physical illness)

Deepfake can be used to ask unanswered questions. This subtheme showed how bereaved people feel that through the use of the deepfake, they can ask the questions they always wanted to ask. It is not strictly about expecting answers, but rather about having the ability to ask questions. Participants considered this valuable, even though they were aware that the answers they received are not from the deceased.

The possibility of having a conversation and asking questions (but you never know if the answer is the truth from the deceased). (Woman, aged 20–29, lost a sibling due to an accident)

Deepfake can be used to get answers to pressing questions. This subtheme differs from the previous subtheme in that participants expressed how they expect to receive answers from the deepfake rather than having the opportunity to ask questions. Many participants expressed a desire to receive answers or, at the very least, felt as if they had, considering that these answers might not reflect what the bereaved would have said.

Hopefully, to receive answers or feel as if I have. (Woman, aged 70–79, lost a partner due to physical illness)

The use of deepfake can give me the chance to get in touch with the deceased. This main theme describes how bereaved people thought they could use the deepfake to connect with the deceased. It encompasses 90 quotations, organised into five distinct subthemes.

Deepfake can be used to reconnect with the deceased. This subtheme describes how reconnecting with the deceased through the deepfake could help the grieving process. Bereaved people mentioned how they could, using the deepfake, communicate with their loved ones and how this could provide temporary relief from the idea of never being able to communicate with them again.

The idea that you can talk to and see your loved one again, and the temporary (?) comfort that comes from that. (Women, aged 50–59, lost a parent due to physical illness)

Deepfake gives the opportunity to see the deceased again in moving images. This subtheme describes how bereaved people valued the idea of being able to see their deceased loved ones again. They felt it could help them open up more easily when the image is moving instead of a non-moving picture. Others expressed that it could help them keep memories of their loved ones alive so they would not forget them. Two participants saw the value in using the deepfake to show moving images of their partner to their children.

To be able to show my children moving images of their father to keep his memory alive. (Woman, aged 40–49, lost a partner due to physical illness)

Deepfake gives the opportunity to hear the voice of the deceased again and talk to them. This subtheme describes how bereaved people desire to talk to their deceased loved one. They believed this could help them remember their loved ones, feel their presence, or help them to slowly say goodbye.

You can ‘talk’ to the deceased person again. I think this might help you gradually let go, instead of suddenly losing the voice of your loved one. This could potentially make the grieving process a bit easier. (Man, aged 20–29, lost a grandparent due to suicide)

The use of deepfake can help in the grieving process. This main theme describes how bereaved people felt the deepfake could, in general, help themselves or others in the grieving process. It encompasses 60 quotations, organised into five different subthemes.

The use of deepfake aids the grieving process. This subtheme includes bereaved people stating more generally that deepfake technology could contribute positively to grief processing without pointing to specific advantages.

Everything that can help must be used. (Woman, aged 70–79, lost a partner due to physical illness)

The use of deepfake can help people with prolonged grief. This subtheme includes participants of whom some do not reflect on their personal experience, but rather on how others (i.e. those with prolonged grief) could benefit from deepfakes. They mention how deepfake can be used to move forward for those stuck in the grieving process.

Still being able to discuss unresolved questions or emotions to overcome stagnation in the grieving process or to transform certain negative thoughts or feelings, such as guilt. (Woman, aged 40–49, lost a parent due to physical illness)

3.5. Themes for the reasons for not using deepfake technology in grief treatment by bereaved people

Table 4 displays the top three main themes and its corresponding most frequently mentioned subthemes for not using deepfake technology in grief treatment by bereaved people. In some cases, less than three subthemes are included due to equal frequency of occurrence.

The use of deepfake is fake and artificial. This main theme describes how bereaved people consider the deepfake to be fake or too artificial, as it might not match the deceased's voice, personality, or image. Bereaved people emphasised that the deepfake is not their deceased loved one and cannot truly give them the answers they might seek. Bereaved people see it as a way of deceiving themselves and find it unnatural. This main theme encompassed 125 quotations, with a total of 14 different subthemes.

The deepfake is fake. In this subtheme, bereaved people noted that their awareness of its fakeness is enough for them to conclude that using deepfake cannot have any positive effects on the grieving process. Some participants argued that this fakeness of the deepfake could hurt the grieving process as it might give the wrong perception of the deceased.

It's not real. What meaning can a bereaved person truly derive from the possible responses of the deceased? I find that a fascinating and complex question. (Woman, aged 50–59, lost a partner due to physical illness)

The deepfake’s answers are not truly from the deceased. Bereaved people mentioned that these fake answers could complicate the grieving process, for instance by distorting your image of the deceased. Several other participants were worried about how these answers would be obtained and who would have control over them.

You might make it more complicated for yourself by receiving answers that are not really from the deceased. (Woman, aged 30–39, lost a partner due to a medical failure)

The deepfake is not really the deceased. This subtheme describes the general sentiment among participants that the deepfake is not the actual deceased. Consequently, the deepfake cannot have the same answers or personality as the deceased had and can, therefore, have a potentially harmful effect on the bereaved. This fakeness makes the use of deepfake ineffective, as no one would believe they are actually talking to their deceased loved ones.

You’re not actually talking to the deceased person … so they can’t have the same personality, which could be negative for bereaved people using this technology. (Man, aged 20–29, lost a grandparent due to suicide)

The use of deepfakes hinders the grieving process. This main theme reflects how participants considered the potential for deepfakes to hinder the grieving process. This main theme encompasses 100 quotations with 10 different subthemes.

The use of deepfake is too confronting. Many bereaved people wrote that it could be too confronting to see their loved ones moving and talking again; they are afraid of the pain and the grief that might come with it. Participants mentioned that the use could thus be potentially harmful to the grieving process and would not have the desired effect on the bereaved.

It can be too confronting to see your loved one; it might actually intensify the grief because after seeing them, you might miss them even more. (Woman, aged 40–49, lost a parent due to physical illness)

The use of deepfake makes accepting the death of the deceased more complicated. Bereaved people wrote that it might delay or even prevent people from accepting the reality of loss, which is considered an essential step in the grieving process.

Confusion about death, you do your best to accept that it is final, and suddenly they are there again. (Woman, aged 50–59, lost a partner due to physical illness)

The use of deepfake entails psychological risks. This main theme entails the potential psychological risks that participants considered when using deepfake technology. This main theme encompassed 83 quotations with 11 different subthemes.

The use of deepfake can be addictive, as it creates a desire to keep talking to the deceased. This subtheme shows how participants considered the use of deepfake to potentially have an addictive effect. They feel that the use would not be a one-time thing, and that bereaved people would constantly come back to talk to the deceased. Participants considered how it could lead to obsession with the deepfake and could be used to not let go of the deceased or to avoid accepting reality and thus impede the grieving process.

Perhaps there’s a risk that you’ll want to repeat it over and over … a kind of dependency that prevents you from letting go of the deceased? (Woman, 60–69, lost a partner due to Physical Illness)

The use of deepfake may cause confusion about what is real and what is not. This confusion among bereaved people may include mistaking the deepfake for being real or difficulties in distinguishing reality from fiction. This confusion about what is real and what is not could result in difficulties in accepting the reality of the loss, which may complicate the grief process for bereaved people.

Not being able to distinguish between reality and fiction. (Woman, 50–59, lost a parent-in-law due to physical illness)

The use of deepfake can intensify grief symptoms. In this subtheme, participants reflected on the possibility that the use of deepfakes could lead to increased grief reactions. Participants expressed how they could see how its use may lead to potential problems with sleeping and general mental health. They considered that it might do more harm than good.

That it can stir up more than it helps. (Woman, aged 50–59, lost a partner due to physical illness)

3.6. Themes for the reasons for using deepfake technology in grief treatment by MHCWs

Table 5 displays the three most frequently mentioned main themes and their corresponding subthemes for utilising deepfake technology in grief treatment, as reported by MHCWs.

Table 5.

Main themes and subthemes in favour vs against using deepfake technology in grief treatment as mentioned by MHCWs (N = 126).

Reasons in favour of using deepfake technology in grief treatment
Main themes (number of occurrences)	Subthemes (number of participants referring to subtheme)
Deepfake technology can be used to enhance existing therapeutic techniques. (124)	Deepfake can be used to address unfinished business. (37) Deepfake can be used to strengthen exposure therapy. (31) Deepfake can be used to help target avoidance. (24)
Pragmatic and general reasons to use deepfake technology. (79)	It is important for therapists to keep up with new technologies. (10) Deepfake can be used to treat prolonged grief. (10) Deepfake can help clients process their loss better and more gradually. (9)
Deepfake can be used to help clients to say goodbye to their loved ones. (56)	The use of deepfake can give clients the opportunity to find closure in the grieving process. (25) The use of deepfake can give clients the opportunity to say goodbye to the deceased. (25) Deepfake can be used to give the client permission from the deceased to move forward with life. (4)
Reasons against using deepfake technology in grief treatment
Main themes (number of occurrences)	Subthemes (number of participants referring to subtheme)
The use of deepfake technology can pose psychological risks. (91)	The use of deepfake could potentially be addictive for the client. (27) The use of deepfake can distort reality. (24) Clients may become attached to the deepfake. (12)
The use of deepfake technology can have a negative impact on the grieving process. (90)	Deepfake can reinforce avoidance of the reality of the loss. (26) The use of deepfake hinders letting go of the deceased. (12) Deepfake can reinforce the denial of the death. (10)a In general, the use of deepfake hinders the grieving process. (10)a
Conducting a treatment with deepfake technology can be challenging and may compromise the client's safety. (58)	The deepfake could potentially say things that are not helpful in the grieving process. (16) The use of deepfake is artificial. (10) It is possible that the deepfake material could be abused. (6)a How is it determined what the deepfake should say? (6)a

^aNot mentioned in text due to equal frequency of occurrence.

Deepfake technology can be used to enhance existing therapeutic techniques. This main theme describes the MHCWs’ mentions of specific therapeutic techniques that could be strengthened using deepfake technology. It encompasses 125 quotations, organised into 10 distinct subthemes.

Deepfake can be used to address unfinished business. Many MHCWs expressed how they saw the opportunity for deepfake technology to allow clients to say the things they always wanted to say to their deceased. This can help clients in the grieving process and resolve unfinished business. Through the use of deepfake MHCWs, bereaved people can resolve this unfinished business in the most realistic way possible. This may help people to move forward.

Being able to say things that were left unsaid to the deceased (e.g. saying goodbye, I'm sorry, I love you). (Woman, 40–49, licensed mental healthcare psychologist, with over 20 years’ experience, and experience with grief treatment)

Deepfake technology can strengthen exposure therapy. MHCWs feel that it can create a realistic addition to exposure therapy. MHCWs were convinced that using deepfake may help bereaved people by providing a very realistic confrontation with the deceased and thus target avoidance behaviour.

Exposure is part of the treatment, and through deepfake technology, the client is exposed to the most realistic possible interaction with a deceased loved one. (Man, aged 20–29, licensed psychologist, with less than 5 years’ experience, and experience with grief treatment)

Deepfake can be used to help target avoidance. In this subtheme, MHCWs considered the potential for deepfake technology to break through clients’ avoidance behaviour. MHCWs expressed that by using deepfake, clients have less of a chance to avoid their emotions. Instead, they may be more inclined to actively and emotionally engage in confronting distressing thoughts, memories, or situations. Furthermore, clients’ expectations before seeing the deepfake could aid in this effect, leading to less room for avoidance behaviour overall.

It may help the bereaved truly feel their emotions when seeing the deceased, making avoidance less possible. (Woman, aged 20–29, licensed psychologist, with less than 5 years’ experience and no experience with grief treatment)

Pragmatic and general reasons to use deepfake technology. This main theme encompassed 79 quotations with 15 different subthemes. It describes the pragmatic benefits and general rationale for using deepfake technology in grief treatment.

It is important for a therapist to keep up with new technologies. MHCWs emphasised the importance of staying up to date with new technologies and treatment options. They considered that the technology will be applied regardless of empirical evidence. MHCWs realised that development and dependency on technology are part of our society and that keeping up with new possibilities is essential, even in therapy. According to MHCWs, keeping up with technology is beneficial and crucial for preventing the mental healthcare sector from falling behind in development.

I believe this technology will be applied anyway, so it’s better for us to take the lead from a scientifically and ethically responsible perspective. (Woman, aged 40–49, licensed mental health care psychologist, with 11–15 Years experience, and experience with grief treatment)

Deepfake can be used to treat prolonged grief. MHCWS considered how the use of deepfake can be of great benefit when dealing with prolonged grief. The use of deepfake technology could be the extra push needed for some people to take that extra step when stuck in their grief process.

It aligns with the current theoretical framework, which assumes that grief can become stuck when the bereaved either excessively or insufficiently allow emotions and thoughts about the loss. (Woman, aged 30–39, licensed psychologist, with 5–10 years’ experience, and experience with grief treatment)

Deepfake can help clients process their loss better and more gradually. MHCWs considered how the use of deepfake technology could generally aid clients in processing their loss (not specifically focused on prolonged grief). It could allow clients to accept their loss, facilitating a gradual overall process. Overall, it can lead to a more manageable and bearable grief process, creating more space for the client's recovery and healing.

It could help clients process the loss more effectively. (Woman, aged 30–39, licensed mental healthcare psychologist, with 10–15 years of experience and experience with grief treatment)

Deepfake can be used to help clients to say goodbye to their loved ones. This main theme encompassed 56 quotations with four different subthemes.

The use of deepfake can give clients the opportunity to find closure in the grieving process. MHCWs expressed that deepfake technology could be used to provide closure. By having contact with the deceased one last time, the bereaved can express how they feel or say the things they would have wanted to say.

People who are left with unanswered questions or the memory of a difficult final moment may, in this way, find the closure they need. (Woman, aged 20–29, licensed psychologist, with less than 5 years of experience and experience with grief treatment)

The use of deepfake can give clients the opportunity to say goodbye to the deceased. MHCWs expressed the potential of deepfake technology for bereaved people to say goodbye to their loved ones. Many bereaved people have never had the opportunity to say goodbye, but now they can finally do so with the help of this technology.

To say goodbye when, for whatever reason, this wasn’t possible. (Woman, aged 30–39, licensed mental health psychologist, with 5–10 years’ experience, and experience with grief treatment)

Deepfake can be used to give the client permission from the deceased to move forward with life. MHCWs noted that many bereaved people can feel reluctant to continue life without the deceased. Through this technology, the bereaved can get permission from the deepfake of the deceased to move forward with life. Some also stated that this could be the final push people might need to take the next steps in continuing their lives without the deceased.

It could be beneficial for the treatment if, in agreement with the client, the deceased could, in a way, expressed that the client is allowed to move on with their life. People sometimes need permission to take steps forward. (Woman, aged 30–39, licensed psychologist, with 5–10 years’ experience, and experience with grief treatment)

3.7. Themes for the reasons for not using deepfake technology in grief treatment by MHCWs

Table 5 displays the top three main themes and the corresponding most frequently mentioned subthemes for reasons for not using deepfake technology in grief treatment as mentioned by MHCWs. In some cases, less than three subthemes are included due to equal frequency of occurrence.

The use of deepfake technology can pose psychological risks. This main theme encompassed 91 quotations with seven different subthemes, reflecting MHCWs’ concerns about the potential psychological risks associated with using deepfake technology.

The use of deepfake could potentially be addictive for the client. MHCWs expressed their concern that using deepfake technology could become addictive. Clients might want to keep coming back to the deepfake and continue talking to their loved ones instead of processing their grief. MHCWs considered that the potential to see a loved one via a deepfake would never be enough. Clients may want to return, resulting in avoiding the reality of the loss. Some stated they might not want to go without the deepfake ever again.

It could potentially be addictive, in the sense that a grieving person may want to maintain the virtual contact and feel ‘content’ with that instead of processing their grief. (Man, aged 50–59, licenced clinical psychologist with more than 20 years’ experience, and experience with grief treatment)

The use of deepfake can distort reality. MHCWs expressed concerns about the clients’ potentially confusing reality and fiction in this subtheme. Clients might perceive the deepfake as real, leading to a reduced grip on reality and promoting delusion among clients. As a consequence, the grieving process would be hindered and could lead to the client losing touch with reality.

I can imagine that some people may find it difficult to distinguish between reality and deepfake. This could potentially lead to a form of dependency on the deepfake, since that’s where the deceased ‘still lives’. (Woman aged 30–39, a licensed mental healthcare psychologist with 11–15 years’ experience, and experience with grief treatment)

Clients may become attached to the deepfake. In this subtheme, MHCWs expressed their concern about the potential for clients to become too attached to the deepfake. They considered the possibility that the bereaved cling to the images and sounds the deepfake creates, and as a result, may find it difficult to let go of the deceased. Other MHCWs emphasised the potential that clients might not develop normal coping strategies, and that they use the deepfake as a comfort tool, effectively becoming attached and dependent on it.

It provides something new to cling to. In grief, you desperately search for the last straws, and now you're given a significant and almost tangible one – only to have to let go of it again. (Man, aged 50–59, psychosocial therapist, with 11–15 years’ experience, and experience with grief treatment)

The use of deepfake technology can have a negative impact on the grieving process. This main theme encompassed 90 quotations with nine different subthemes. It covers the MHCWs’ perspective on the negative impact deepfake technology can have on the grieving process.

Deepfake can reinforce avoidance of the reality of loss. MHCWs expressed their concerns that the application of deepfake technology could reinforce avoidance of the reality of loss, making it harder to accept the finality of death. In interacting with a deepfake, clients could experience a false sense of the presence of the deceased while being aware that they are dead. This could delay or obstruct the process of realising that their loved one is truly gone and will not come back. Other MHCWs agreed and expressed that utilising deepfake technology could increase or even cause prolonged grief among clients. The client might feel as if the deceased is still there on an emotional level, preventing them from moving forwards.

It conflicts with the fundamental finding from research on prolonged grief: when the bereaved do not fully acknowledge that the death of their loved one is truly irreversible. (Woman, aged 30–39, licensed mental health care psychologist, with 11–15 years’ experience, and experience with grief treatment)

The use of deepfake hinders letting go of the deceased. This subtheme covers the MHCWs’ position on using deepfake and how it could prevent clients from letting go of the deceased. Clients may become emotionally attached to the deepfake and may struggle to let go or move on with their lives. This could result in the grieving process being prolonged, as clients find it difficult to separate from the deepfake of their loved ones.

I believe that a grieving process requires gradually letting go of someone (or holding on in a different way) in order to move forward with one's own life. Deepfake technology could have the opposite effect and prolong the grief instead. (Woman, aged 50–59, grief counsellor, with 5–10 years’ experience, and experience with grief treatment)

Conducting a treatment with deepfake technology can be challenging and may compromise the client's safety. This main theme encompassed 58 quotations with 11 different subthemes. It describes the potential practical pitfalls when applying deepfake technology in grief treatment and how this can affect the grieving process. Additionally, the theme addresses the threat of data leaks and the potential abuse of deepfake technology.

The deepfake could potentially say things that are not helpful in the grieving process. MHCWs expressed worries about the things the deepfake might say in the treatment. Several MHCWs questioned who is responsible for constructing these answers and who is controlling the deepfake. MHCWs considered the possibility that the things the deepfake says might not align with how the deceased was. This could consequently lead to the bereaved becoming confused, shocked, or even traumatised. MHCWs stressed caution when using deepfake, given that grief could be worsened. Furthermore, MHCWs wondered if the deepfake were to say something wrong, it would be possible to repair the damage it may have caused.

Who determines what the deceased person's response is? This could have a negative effect if the wrong thing is said. (Woman aged 30–39, licensed mental healthcare psychologist, with 5–10 years’ experience, and without experience with grief treatment)

The use of deepfake is artificial. In this subtheme, MHCWs emphasised that the artificiality of the deepfake could have adverse treatment effects. Some considered that the deepfake could misrepresent the deceased, causing anger and frustration in the bereaved. The deepfake might not meet the expectations of the bereaved and not align with their memories of the deceased. Some MHCWs suggested that this artificiality of the deepfake could result in the entire treatment being ineffective. It could even damage the memories and perceptions the bereaved have of the deceased.

It remains an imitation of appearance, actions, voice, and language use that differs from the memory of the deceased. (Woman, aged 30–39, licensed mental health care psychologist, with 16–20 years’ experience, and with experience with grief treatment)

4. Discussion

This study explored how 159 bereaved people and 126 MHCWs perceive the potential value and risk of using deepfake technology in grief treatment. In total, 749 textual responses from bereaved people and 654 from MHCWs were qualitatively analysed. About one out of four bereaved people and one out of three MHCWs indicated that they were willing to use deepfake technology in grief treatment if given access.

4.1. Deepfake technology in grief treatment to target unfinished business

Our primary finding was that bereaved people and MHCWs considered deepfake technology potentially valuable to address unexpressed or unresolved issues between the bereaved person and the deceased, which is also defined as ‘unfinished business’ (Holland et al., 2018). Bereaved people often expressed the desire to say things that were left unsaid, such as unexpressed feelings, questions, or relationship issues, which are especially relevant after sudden or violent loss (Klingspon et al., 2015).

Similarly, MHCWs acknowledged that unfinished business is frequently targeted in grief treatment with techniques, such as letter writing, imaginal exposure, or imaginary rescripting. MHCWs expressed the potential for deepfake to be used as a helpful tool to help resolve this unfinished business, especially for people struggling with emotional avoidance. This suggests that deepfake technology may serve as a technological tool to enhance current evidence-based grief interventions for people with PGD, such as the ‘empty chair method’, or imaginal exposure, especially for clients who struggle with imagination (Jordan & Neimeyer, 2003; Shear, 2015; van Minnen et al., 2022). Mechanistically, this positions deepfake as a therapist-guided tool that may make imaginal dialogue and imagery-based techniques more concrete by externalising the conversational ‘other’ and providing richer cues to support engagement with avoided loss-related material (Shear, 2010).

4.2. Contrasting perspectives of reconnection with the deceased

A second main finding revealed a conflicting perspective between bereaved people and MHCWs about using deepfake technology in grief treatment. Bereaved people often expressed a wish to ‘see’ or ‘speak with’ the deceased, considering this ‘reconnection’ with the deceased as potentially comforting and meaningful. However, MHCWs recognised this simulated reconnection as clinically risky and potentially countertherapeutic. MHCWs argued that simulating reconnection with the deceased could hinder facing the reality, or irreversibility, of the loss, which is one of the main goals of cognitive behavioural therapy, which is the first-choice treatment for PGD treatment (Boelen et al, 2006; Boelen et al., 2021; Reitsma et al., 2023; Shear, 2010). This contrast can be interpreted in light of continuing-bonds perspectives, where ongoing connection may be experienced as supportive, but may also become problematic depending on how the bond is maintained (Eisma & Nguyen, 2023; Field & Filanosky, 2009). More specifically, externally expressed continuing bonds (e.g. illusions or hallucinations of the deceased) have been associated with post-loss distress, whereas internally expressed continuing bonds (e.g. beliefs that the deceased communicates through dreams) were not associated with post-loss distress (Field et al., 2013; Field & Filanosky, 2009).

It seems therefore critically important that deepfake technology is not used as a tool to simulate reconnection with the deceased. Because an unbounded, unguided, or stand-alone simulation may externalise the bond with the deceased in ways that reinforce avoidance of the reality of the loss (Eisma & Lenferink, 2023; Field & Filanosky, 2009). This co-occurrence of approach and avoidance behaviours, also referred to as the approach-avoidance hypothesis, appears to be characteristic to people with PGD (Eisma & Lenferink, 2023). If deepfake technology is considered to be used in bereavement care, psychoeducation is of utmost importance regarding the potential usefulness and limitations of deepfake technology. However, it is of utmost importance that the concerns raised in both groups are examined in scientific research before considering its adoption in clinical practice (Hoek et al., 2025).

4.3. Ethical concerns

Both groups highlighted the potential for harm when simulating a deceased person's image, voice, or behaviour. Bereaved people expressed concerns about the simulation being perceived as too artificial or inconsistent with their memory of the deceased. These reflections align with broader critiques of grief bots and posthumous AI, where authors highlight the moral ambiguity and potential dissonance created when simulated interaction deviates too far from memory or expectation (Hollanek & Nowaczyk-Basińska, 2024). MHCWs mainly expressed concerns regarding the lack of consent of the deceased, the risk of misrepresenting the deceased person, possible misuse of deepfake materials, and misleading bereaved people with simulations.

The concerns raised align with existing literature, which highlights the risks of simulating the dead without proper boundaries or involvement from next of kin (Hoek et al., 2025; Hollanek & Nowaczyk-Basińska, 2024). The clinical implementation of deepfake-based interventions would require clear protocols co-developed by clinicians, ethicists, and bereaved families, and evaluated in scientific research, before implementation in clinical practice is considered, to ensure the respectful and safe application of deepfake technology in bereavement care.

4.4. Strengths and limitations

One of the key strengths is the dual qualitative approach, which captures the opinions of both bereaved people and MHCWs. This approach formed a nuanced understanding of the potential usefulness and nonsense of using deepfake technology in grief treatment. Second, participants were recruited through materials titled: ‘The usefulness and nonsense of using deepfake technology in grief treatment’. This likely contributed to a sample including people from both sides of the argument.

Several limitations should be noted. First, the participants only reflected on a hypothetical situation rather than a lived experience with a deepfake intervention, which can lead to an overestimation or underestimation of its impact and real-world application. Second, recruitment of bereaved people took place via a website about grief treatment and social media, likely introducing self-selection bias towards people who are digitally engaged and actively seeking grief-related information. This limits the generalisability to all bereaved people. Additionally, the MHCWs sample was convenience-based, relatively experienced with treating trauma and grief disorders, predominantly female, and highly educated. The latter two are not surprising as most worked in the Netherlands where a university degree is needed to work as a licensed psychologist and of whom the majority is female (Dutch Ministry of Health, Welfare and Sport, 2026). However, our findings may therefore not generalise to MHCWs with other gender identities or those who work in domains other than trauma and grief. Third, generalising our findings to other countries may be limited, as this study reflects the Dutch context where mental healthcare is reimbursed, but deepfake-based services (e.g. Deeptherapy) are not, and where access and perceptions may differ from countries with more limited care or no such services. Fourth, findings should be interpreted as Dutch stakeholder perceptions after viewing a short videoclip of a deepfake session, rather than evidence of clinical effectiveness. Fifth, when analysing the data, we distinguished participant-derived terms from established psychological constructs, noting that overlap (e.g. avoidance, dependency, exposure and acceptance) can cause misinterpretation across everyday, stakeholder, and clinical contexts. Future research should define, operationalise, and examine these constructs’ potential interpretive discrepancies. Furthermore, future research should examine what factors hinder or benefit the willingness in using deepfake-technology in grief treatment from the perspective of bereaved people and MHCWs. This allows for examination of profiles of people who may be more or less receptive to such interventions (cf. Lopez-Castroman et al., 2021).

5. Conclusion

This study provides a preliminary account of how bereaved people and MHCWs perceive the potential usefulness and nonsense of deepfake technology in grief treatment. While bereaved people and MHCWs emphasised the emotional value of it, particularly in addressing unfinished business, MHCWs also reflected more critically on therapeutic risks, ethical concerns, and clinical boundaries. Deepfake may hold promise as a technological adjunct tool to evidence-based interventions in specific contexts, such as PGD treatment targeting unfinished business, but only under carefully controlled conditions, considering contraindications, such as risk for strong emotional dependency and dissolution tendencies. As digital tools increasingly enter the realm of mental health care, this study opens a crucial dialogue on how new technologies might not only reshape the therapeutic process, but challenge our most fundamental understandings of presence, loss, and human connection.

Appendices.

Appendix 1. Introduction to deepfake technology in grief treatment

You are about to watch a short segment from a documentary. This documentary, released in 2020, showcases an experiment where bereaved people engage in an interactive video conversation with their deceased loved ones using deepfake technology. In grief treatment, deepfake technology uses a computer to create realistic but artificial videos of the deceased. By manipulating an existing photo or short video of the deceased, deepfake technology slightly moves the head, mouth, and eyes. In the documentary, manipulated images of the deceased are synchronised with a voice actor's mouth movements, with the actor playing the role of the deceased. The video features Ron and Patricia, who are reconnected with their deceased daughter.

Deepfake technology could serve as a tool to support existing grief treatment techniques. In grief treatment, people are sometimes encouraged to write a letter to the deceased. In addition to writing the letter, deepfake technology could be used to have the letter read to the deceased in a video. Grief treatment is aimed at people who experience intense and prolonged grief symptoms after a loss. These people face challenges in daily life and, therefore, seek professional psychological support.

Transcript³ of a 27 s videoclip from a Dutch documentary ( Nejal, 2020 ).

[Video begins]

Two people, a middle-aged man and woman, sit closely together indoors. Both wear headphones and look at a laptop screen in front of them. Their posture is upright and attentive; they both smile. The woman says ‘Hey’ in an uplifted tone.

[On screen: deepfake animation]

A girl appears, seated outdoors in a bright, green garden. She looks directly into the camera with a soft, calm expression. She smiles slightly.

While viewing the deepfake animation, the man says in Dutch: ‘Hey darling.’

[Cut to another woman]

The woman is also sitting indoors and wearing headphones, looking at a laptop screen in front of her. She is smiling and crying at the same time.

Dutch subtitles on the screen state: ‘Using a new technique, bereaved people can talk with their deceased loved ones’.

[Cut to another man]

An elderly man wearing glasses and headphones watches a laptop screen. His face is focused and slightly tense, but he is also smiling slightly.

[Return to the middle aged man and woman]

The woman says in Dutch: ‘Do you regret that you did this?’

Her facial expression changes. Her eyes narrow slightly, and her mouth tightens. She appears emotionally affected.

While still viewing the couple, the deepfake animation of the girl says in a gentle and reassuring tone in Dutch: ‘I regret that I could not give you what you wanted. You really wanted to help me’.

The middle-aged woman begins to cry.

She raises her hand to wipe her eyes as she continues watching the screen.

The man beside her lowers his gaze briefly, appearing quiet and reflective, then looks back at the screen.

[Return to the other woman looking at the laptop screen]

The woman is staring down. A hand closes the laptop in front of her.

[Final shot]

The screen fades to black.

Appendix 2. Data cleaning details

A total of 168 bereaved people completed the survey. We excluded participants whose loss occurred <12 months earlier (N = 6), whose open-ended question responses were missing (N = 2), and who did not watch the video segment (N = 1). This led to a final sample of 159 participants.

In total, 130 MHCW completed the survey. Two participants had too much missing data and two had no clients or active treatment hours. Therefore, the final sample included 126 MHCWs.

Appendix 3. Detailed iterative three-stage coding process

1. Familiarisation with the data and interpretation: In the first round, the reasons for using deepfake technology were thoroughly reviewed to familiarise with the data and context. During this stage, the data were broken down into manageable ‘units’ of meaning to create a more comprehensive understanding of the data. This round helped build an initial framework for subsequent coding.

2. Initial coding: In the second round, open coding was applied, a method in which codes are developed directly from the content of the reasons without using a predefined set of codes.⁴ Each reason was segmented into one or more units, which were then labelled with a subtheme depending on its content. Units expressing similar ideas were grouped under the same subtheme. For example, all units mentioning how they could help with exposure were now labelled ‘Exposure’.

3. Code refinement: In the third coding round, all units were revised to identify subthemes and eliminate overlap and redundancies. When multiple subthemes covered similar content, they were merged to create more comprehensive subthemes. The final codebook included a list of all subthemes and their corresponding frequencies of occurrence.

Appendix 4. A qualitative coding example

Category	Units	Subthemes	Main themes
Bereaved people
Reason in favour	Expressing unspoken things. Being able to say sorry because I missed his last phone call.	Deepfake can be used to express things left unsaid to the deceased.	The use of deepfake can provide the opportunity to address unfinished business.
Reason against	Confusion about death: while you are doing your utmost to accept that it is final, they suddenly appear again.	The use of deepfake makes accepting the death of the deceased more complicated.	The use of deepfake hinders the grieving process.
Mental health care workers (MHCWs)
Reason in favour	Further strengthening the effect of, for example, imagery rescripting.	Deepfake can be used to reinforce rescripting.	Deepfake technology can be used to enhance existing therapeutic techniques.
Reason against	When reality and artificiality become too intertwined.	The use of deepfake can distort reality.	The use of deepfake technology can pose psychological risks.

Appendix 5. Work experience characteristics of MHCWs

Characteristics of MHCWs (N = 126)
Job function, N (%)  Master student  PhD candidate  Social worker  Counsellor  Licensed psychotherapist  Licensed psychologist  Licensed mental health care psychologist  Licensed clinical psychologist  Licensed general practice mental health professional  Licensed psychiatrist  Other	2 2 1 2 6 42 36 12 3 2 18	1.60 1.60 0.80 1.60 4.80 33.30 28.60 9.50 2.40 1.60 14.30
Workplace, N (%)  Owner and practitioner of own private practice for therapists/psychologists without staff Owner and practitioner of own private practice for therapists/psychologists with staff Practitioner at a small mental healthcare institution (0–49 employees) Practitioner at a large mental healthcare institution (50–100 employees) Combination of the options described above Other	11 1 10 79 9 16 0	8.70 0.80 7.90 62.70 7.10 12.70 0
Number of years employed in mental healthcare, N (%)  Less than 5 years  5–10 years  11–15 years  16–20 years  More than 20 years	35 32 21 12 26	27.80 25.40 16.70 9.50 20.60
Number of clients treated (per week), M (SD)	16.45	6.83
Number of work hours (per week), N (%)  Less than 8 h  8–16 h  16–24 h  24–32 h  32–40 h  More than 40 h	7 24 32 46 17 0	5.60 19.00 25.40 36.50 13.50 0
Most often treated mental disorders, N (%)  Trauma- and stressor-related disorders (e.g. PTSD)  Depressive disorders  Somatic symptom disorder and related disorders  Anxiety disorders (e.g. specific phobias and panic disorders)  Personality disorders  Neurodevelopmental disorders (e.g. ASD, ADHD)  Obsessive-compulsive and related disorders  Disruptive, impulsive-control, and conduct disorders  Eating disorders  Substance use disorders  Bipolar and related disorders  Neurocognitive disorders  Dissociative disorders  Schizophrenia spectrum and other psychotic disorders  Other	42 19 18 16 10 5 3 1 1 1 0 0 0 0 10	33.33 15.10 14.30 12.70 7.90 4.0 2.4 0.80 0.80 0.80 0 0 0 0 7.90

Note: M = mean, SD = standard deviation.

Funding Statement

This work was supported by the University of Twente.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The pseudononymised qualitative data and codebook used in this study are available via this link: .