Restorative Deepfake Intervention Netherlands (REDEFINE):
Deepfake survivor-perpetrator confrontation for young adults with sexual violence-related PTSD

Sexual violence is a common potentially traumatic experience that may have
major psychosocial consequences. In 2020, 11% of the Dutch population indicated
having been exposed to sexual violence in the previous year. Among the victims
are a disproportionate number of young people: 47% of women and 13% of men aged
between 18 to 24 have experienced some form of sexual violence.
Exposure to sexual violence is a strong risk factor for developing
posttraumatic stress disorder (PTSD). The recommended treatment for PTSD
consists of individual trauma-focused therapy, such as EMDR therapy or
prolonged exposure. Treatment leads to a clinical improvement in around 54% of
cases. However, even after effective treatment, 60% of patients continue to
have residual complaints. Negative moral emotions (guilt, shame, anger) and
negative posttraumatic cognitions about self and others are seen as negative
predictors for a positive treatment response.
To reduce the negative moral emotions and negative cognitions of patients with
PTSD after sexual violence, a victim-perpetrator confrontation may be helpful.
For the victim, the aim of such a confrontation is to achieve psychosocial
recovery by discussing the impact of the act, asking questions, and/or
receiving an apology. Research shows that victim-perpetrator confrontations
lead to, among other things, increased feelings of justice and empowerment;
increased self-confidence and trust in others; reduced negative feelings of
anger, fear and guilt; a reduction of negative cognitions about the world;
reduced fear of revictimization; and reduced desire for revenge.
Victim-perpetrator confrontations also reduce the severity of the core symptoms
of PTSD (reexperiencing, avoidance and excessive arousal) by up to 40%.
Although real-life victim-perpetrator confrontations can help reduce PTSD
symptoms in victims, it is not always possible to have an actual confrontation
with a perpetrator. Not all victims and perpetrators are willing or able to do
this. In that case, technological innovation offers opportunities to conduct
such a confrontation virtually.
In a deepfake victim-perpetrator confrontation, the victim has a Zoom call with
their 'perpetrator'. A photo of the perpetrator is converted into a realistic,
repeating 9-second GIF, and the perpetrator's words are spoken by a therapist
who sits in another room behind the screen. The technology for this was
developed and facilitated by the University of Amsterdam/3DUniversum via an
online platform. This intervention is experimental, and scientific evidence
currently consists of two promising case studies. In these case studies the
intervention was assessed as realistic and positive. There was also an increase
in positive emotions, dominance in relation to the perpetrator and
self-forgiveness, as well as a decrease in negative emotions, self-blame and
severity of PTSD symptoms.
Given the experimental nature of the intervention, it is necessary to further
examine acceptability and effectiveness. When shown to be acceptable and
effective, the intervention may then be implemented in the treatment of
patients with PTSD related to sexual violence.

To evaluate the acceptability and effectiveness of a deepfake
victim-perpetrator confrontation in young adults with sexual violence-related
PTSD.

The study concerns a non-concurrent multiple baseline single case experimental
design (SCED); with randomization of baseline phase (10, 15 or 20 days), 21-31
daily measures (depending on length of baseline phase) as well as four main
assessments, and 10 participants.

The intervention concerns a deepfake victim-perpetrator confrontation.
During this intervention, a frontal photo of the perpetrator is temporarily
uploaded in a secure online program developed and facilitated by 3DUniversum.
There it is converted into a 9-second GIF (a repeating, lightly moving film)
with a benevolent expression.
Subsequently, the victim can engage, from behind a computer screen, in a
conversation with the 'perpetrator', in other words: with a therapist who is in
another room behind another computer screen. In this conversation, the words of
the 'perpetrator' appear to be spoken by the person portrayed in the GIF but
the voice of the therapist remains unchanged.
During the victim-perpetrator confrontation, the victim confronts the
'perpetrator' with his/her memory of the event and the impact of the event, and
asks questions. The desired content and outcome of the deepfake confrontation
is discussed with the victim during a preparatory session, in which the victim
also provides additional information about the event and the perpetrator. This
ensures an optimal match between the content of the deepfake confrontation and
the aim and experience of the victim.
The deepfake 'perpetrator' may then respond to the victim in a way that is both
truthful and helpful (i.e. most likely to meet the objective of the victim).
Depending on the objective of the victim, remarks may be made such as: 'I hear
what you're saying', 'I didn't realize that before', 'that must have been
hard', 'it wasn't your fault', and so on. These remarks take into account the
potential stance of the real perpetrator, meanng that, for example, the
deepfake 'perpetrator' would not apologize when the actual perpetrator would
not be inclined to apologize. Questions of the victim may be answered based on
the information provided in the preparatory session and based on general
information on sexual violence.
The victim is supported by a second therapist who is present in the same room.
The intervention was developed during a first pilot phase and subsequently
developed in a second round in collaboration with experts and experience
experts (patient representatives).

Participation takes a maximum of 585 minutes (9:45 hours), including the
inclusion meeting, preparatory session, deepfake session, four main
assessments, and 21-31 daily assessments.

Participation in the study may be emotionally challenging, because of the
assessments as well as the intervention itself. However, during a pilot phase
the intervention was evaluated as positive. In addition, there was an increase
in positive emotions, dominance in relation to the perpetrator, and
self-forgiveness, as well as a decrease in negative emotions, selfblame and
PTSD symptom severity.