“I Did Watch ‘The Handmaid's Tale’”: Threat Modeling Privacy Post-roe in the United States

2.1 Privacy Research and Technology in the Reproductive Space

When it comes to reproductive health technology (the term we use to broadly refer to apps and IoT that monitor ovulation, pregnancy, breastfeeding, menstrual cycles as well as mood, and other health data from which reproductive status can be gleaned) much focus on HCI has been on the benefits they offer and not their privacy implications. For example, some studies look at how to better support lower-income [25, 77, 78] and diverse populations [89] and those with other pregnancy challenges [8]. But reproductive technology is notorious for privacy vulnerabilities because they deal with medical data but are not regulated by health privacy laws in the US. Like other apps, they share data—in this case, sensitive data about reproductive health—with third parties. For example, a review of 23 of the most popular “women's mHealth apps” found that, among other privacy concerns (like behavior and location tracking), the vast majority share data with third parties [2]. Other recent studies of “fertility tracking apps” [3, 73] (also referred to as “menstruapps” [87]) that track people's reproductive cycle for the purposes of helping them predict their periods highlight how these apps capture more than just period data including, general health, sex, pregnancy, and data about partners, children, and friends [3]; do not treat reproductive data as private; and use questionable tracking practices, for example initiating trackers before users interact with the app [87].

While reproductive technologies are notably not required to comply with the Health Insurance Portability and Accountability Act (HIPAA) which sets standards in the US that protect sensitive patient information, they have nonetheless demonstrated some notable vulnerabilities over the years, making them potential vectors for privacy violation in the post-Roe era. For example, the popular app Glow was the subject of a class action lawsuit in 2016 for major security vulnerabilities that would have allowed someone with no hacking skills to access users’ personal data [10]. More recently, Flo Period and Ovulation Tracker was charged by the Federal Trade Commission with sharing information with third parties like Facebook, Google, and others [110]. Flo's privacy policy also stipulates that it will share information with the police [49].

Reproductive technologies are thus a space where privacy concerns have rarely been front and center when it comes to how humans interact with them, even while concerns about their data flows abound. This speaks, perhaps, to HCI's complacency when it comes to privacy more generally.

2.2 Beyond Reproductive Technology: Other Concerns for Reproductive Privacy

Scholarship about reproductive health and technology beyond reproductive technology has examined social media use in relation to pregnancy journeys [4, 5, 7] including privacy concerns in sharing about stigmatized reproductive health complications [4] as well as pregnancy loss disclosures among LGBTQ individuals [81], and support seeking among peers for pregnancy [53]. A full review of the former is outside the scope of this article. An investigation by Gizmodo in 2022 found that dozens of data brokers in the US sell data on pregnant people [105]. The investigators note that one company was found to be selling access to over 60 million users labeled as “Pregnant and Maternity Life Stage,” while another was offering access to nearly 10 million devices labeled “pregnancy test kit” [105].

According to the Surveillance Technology Oversight project (STOP) “[p]rosecutors will obtain geofence warrants to track those at reproductive health clinics, even clinics out of state. Investigators will use keyword search warrants to identify everyone searching for abortion clinics, abortifacients, and even medically accurate information about abortion care” [24].

Other data includes a “catalogue” of those using birth control [105]—a potential target in some states looking to restrict their use. These catalogs are presumably derived from purchasing information, suggesting that, under some circumstances, one need only use a credit card to reveal reproductive status—no reproductive technology app required. While some investigations note that reproductive technology apps have explicit rights to share their data with law enforcement (e.g., study by Forbes [19]), it is not clear, in a market capitalist system, that any of the companies profiling reproductive-related purchases owe their users a greater debt of privacy. Increasingly, a case can be made that the only way to keep reproductive health information safe is to avoid all use of digital tools and leave devices at home when engaging in activities related to reproductive health [88], which could be a lot of activities. Surveillance by third parties has become so pervasive that it is impossible to turn off or retreat from.

It might be argued that the current remedial approach, emphasizing the risk of specific apps and encouraging “abstinence” from a particular technology is narrowly focused on alleviating the symptomatic evidence of the ailment rather than the disease. The underlying problem is our surveillance capitalist society [31, 108]. With so much at risk and so many threat vectors in play—e.g., most obviously, our Google histories, our unencrypted conversations, our social interactions, our confidential records between healthcare providers, should they be subpoenaed— it is important to examine what those who may become pregnant at risk of reproductive surveillance actually conceive of those risks including the rich troves of data they are supplying to commerce and now, potentially, to legal systems.

2.3 Privacy Research and Literacy in HCI

Privacy research has, understandably, focused on how users behave in a given context—on a platform. A traditional approach to understanding user privacy has been to explore app(s) or platform(s) and how users engage with them—their privacy settings, use, and so on—to gauge exposure to threats and how users are (or are not) attuned to them. Take for example, this early study establishing the privacy paradox (the idea that people act in ways that are at odds with their privacy attitudes or intentions) as a cornerstone of platform-based research exploring how teens were using social networks in ways that exposed them to risk to better understand these contradictions [9]. This framing has led to privacy literacy efforts like how to nudge [1, 100, 103] or persuade users [60]. Research into informal literacy has suggested that informal stories from personal networks have an impact on privacy behaviors [79, 82], though it's not clear the extent to which these are simply people replicating guidance, not in an effort to thwart perceived threats.

Reflecting on engagement with privacy studies of this kind, it is understandable that researchers focused on behaviors in narrow arenas. In fact, Daniel Solove, in criticizing the privacy paradox, in particular, makes the argument that it is simply illogical to look at broad privacy attitudes and compare them with actions in a particular context [91]. Still, the post-Roe world presents an opportunity to reset our approach and consider that privacy research must contend with the Faustian bargain that we may have once been able to afford. But as researchers have pointed out, users now represent a constituency of resignation (i.e., prone to a “sense of helplessness” in the face of routine surveillance by companies) [36]. Indeed, more recent criticism from scholars like Draper and Turow suggest that the privacy paradox is the result of corporations promoting “digital resignation” [36]. According to a report by Turow et al., the more we know about the laws and practices of online marketers the more resigned we are to accept their privacy violations [96].

This study's goal is to examine what types of strategies are being used by a population who are vulnerable to privacy violations because of (recent changes to) reproductive health laws in the US. First, it is not clear that those not experiencing immediate risk are not resigned as Draper and Turow suggest in which case privacy literacy may not be what's needed. Second, it is not clear that literacy efforts are appropriately attuned to the needs of individuals to manage specific, intersectional privacy incursions where the surface area of risk might preclude nudging or stories and require nuanced strategies yet undefined. Privacy literacy research presumes a certain platform-centric focus on privacy settings and not the terrain of risk that might be encountered by someone who must worry about what they say and do across social and digital spaces—what this study explores.

2.4 Surveillance Capitalism and the Reproductive Health Market

What scholars call surveillance capitalism [107, 108] which shares some similarities with data colonialism [30] both include the constant harvesting of personal data that normalizes the futility of “collective public anger” [36]. Surveillance capitalism has enabled decades worth of big data collection and prediction including around pregnancy. Expecting pregnant individuals is an incredibly valuable target market. Predicting their behavior to identify their delivery date down to the month is big business [37]. Keeping this information private requires such a massive effort as to be essentially unsustainable. Indeed, Vertesi found that while abstaining from web searches about expecting was not hard, using social media and purchasing products without leaving a trail—which requires, for example, sticking to cash, burner e-mails, prepaid cards, VPNs, and Amazon mail lockers—was not a supportable privacy practice [58, 97].

While the technology policies and practices that undergird surveillance capitalism are a concern, so too is the growing sense that new laws around, for example, reproductive rights supercharge what Foucault conceptualized as the norm-abiding power of panoptic environments [45], in which the threat of surveillance gets instrumentalized through technology by people and institutions [44] and becomes what Foucault describes as a state of governmentality in which the governed take part in governing others. What role does HCI have when privacy covers terrain well beyond devices, to encompass the offline world? What roles do academics have when the laws prohibit certain activities?

3.1 Participant Recruitment

In September of 2022, we posted calls for participation for our study on Twitter and on Craigslist in the District of Columbia (DC), Oklahoma, and Texas. The call invited people for whom the decision to overturn Roe v. Wade was personally relevant because they are (or could someday become) sexually active in ways that may lead to pregnancy; trying to conceive/are pregnant with or without IVF; and/or have other reproductive-related concerns.

We chose DC to reach a diverse set of participants with respect to race/ethnicity [23] and sexuality [48]. We chose Oklahoma [80] as an example of a state with the most strict abortion ban and Texas where abortion is also outlawed and in which the so-called “bounty law” has created additional complexity to privacy management [13, 61, 109].⁶ We also shared the call with people from our professional and personal networks, asking them to distribute it to people they thought might be interested in participating. Our networks may also have drawn participants from other states where abortion was under threat, but no state laws had officially banned it yet. Ultimately, we got the most participants from Craigslist (see Table 1).

Prospective participants were invited to take a screener survey and submit their e-mail (either their own or a more secure ProtonMail e-mail they were invited to create for the study—we shared the link and told them it was an option). The screener survey confirmed eligibility if respondents said they (1) reside in the US, (2) were between 18 and 49, and (3) were able to become pregnant with or without fertility assistance. We chose only to speak with adults (i.e., 18 and over) because of the sensitive nature of this topic. We set the age limit at 49 because, even while there are individuals getting pregnant beyond the age of 49, most healthcare data about pregnancy stops before 50; we also note that older individuals that could not become pregnant (or for whom the risk of an unwanted pregnancy was not as great) may have mostly shared concerns about others’ risk (e.g., their family members, friends) and not themselves. Additionally, we screened participants for reasons why the Roe decision was personally relevant (notably this was asked to be inclusive of those who had reproductive health risk), any privacy technology they used, age, location, gender, sexuality, race/ethnicity, income, education, and location. We sought a mix of participants to increase the diversity of our sample, particularly making sure to include historically marginalized identities along dimensions of sexuality, race, income, and education.

In all, 199 people took our screener survey from which we reached out to 32 people.⁷ We were able to schedule an interview with 15 participants; those 17 whom we e-mailed but did not interview either did not respond or follow-up.

Our interview sample of 15 participants includes 4 who self-describe as bisexual or queer, 4 as Black/African American, 3 as Asian or Pacific Islander, 3 as Hispanic or Latino/Latina/Latinx, 2 as Middle Eastern with overlap among those categories (see Table 1). Our sample includes one participant whom we are not sure was telling the truth about their identity, and we do not reflect their views except in the very few instances when they seem to emphasize others’ views. We did our best to screen participants with our screener survey and at the beginning of the interviews and, with the exception noted, have no other doubts about participants who risked a lot to share their stories about their reproductive privacy and health.

We told survey respondents that we would reach out to them if they were selected to participate in the interview noting that we were looking for a diverse set of participants. All participants were self-described as women, though the survey was open to everyone who could become pregnant but did not identify as a woman such as transgender men or non-binary individuals. We did not ask for pronouns out of concern that the practice would not be familiar to some of our participants and refer to participants, all of whom identify as women, as “she/they/them.”

3.2 Data Collection

We conducted 15 interviews off-video over Zoom. All interviews were transcribed by Zoom, with additional editing by the first author. The consent and honorarium, as well as some more sensitive conversation, was done off record. The recorded portion of the interviews was an average of 32 minutes and lasted between 13 minutes and 56 minutes. The interviews were a sufficient length to capture participants’ context as it related to reproductive health and Roe as well as privacy technology strategies, and the context behind those decisions. It was also important to us to not gather any more data than we absolutely need because of the sensitive nature of this study within the US context. A third of our interviews where we recorded for 40 minutes to an hour, and in some of those cases, we would have liked to have spoken with participants more but would have well exceeded the 60 minutes requested in our research call. One interview was conducted over both Zoom and other methods (not identified for privacy reasons) because of challenges with the participant's WiFi.

In our interviews, we asked participants to describe why the Supreme Court decision to overturn Roe v. Wade was personally relevant and asked them to describe those circumstances/their reproductive health in the context of how they were sharing this information before and after the Supreme Court decision—asking specifically about changes to who they shared with and what technology they used. Interviews probed on how and why on examples that mostly they provided like period tracking, e-mail, text, social media, health records, and online health portals. Participants were also asked where they went for sources/resources about any changes to their information-sharing behavior; their familiarity with the laws in their state; and should they need an abortion, what would they do?

While we asked about technology use, we structured our conversation to avoid priming participants to think about technology risks. For instance, we asked what, if anything, had changed about their technology use but avoided probing about numerous technologies or devices or asking about it in negative terms.

3.3 Analysis

Following each interview the first author wrote memos, which would later become the vignettes in our findings. The first author conducted a thematic analysis [15, 17] by open coding concepts that would ultimately become the themes described in the findings. Themes represent patterns of codes in the data grouped into domains or concepts. They represent concepts that are most salient, not necessarily most frequently encountered—though we make clear in our findings when themes are based on salience rather than occurrence [16].

3.4 Interpretative Lens and Framing of Results: Intersectionality, Privacy Intermediation, and Use of Thicker Narrative

Intersectionality operates as a framework and methodology. While we approached this research with a phenomenological [85] stance, we employ intersectionality [27–29] as a frame for our analysis because it lends critical importance to thinking about power in relationship to multiple, interconnected social coordinates and vulnerabilities. One core insight of intersectionality is that “conditions of social and political life” are “not shaped by any one factor” but that they build on one another and one must explore these interconnections in relation to power [29]. In our work, the media, authorities, reproductive health workers, people's social and work networks, and so on. all have the power to influence privacy strategies. Understanding what it is individuals do, however, require a sensitivity to the multiple dimensions of experience. By looking at the interdependence of reproductive health risk, power, culture, and other facets of identity we aim at constructing the type of nuanced view that we believe is an important orientation for privacy intermediation—a way of thinking about and ultimately providing holistic and integrative privacy management for those who are vulnerable and which we expand on in the discussion. To that end, while our results were derived using thematic analysis, we use details about participants (“thicker narrative”) to provide some scene-setting and context for understanding nuanced privacy decisions, drawing inspiration from narrative techniques used in ethnography experiences that emphasize telling a “great” story by evoking images and feelings and communicating people's experiences [54].

In taking an intersectional lens to our interpretation, we draw heavily from a black feminist scholar, Patricia Hill Collins. Collins puts forth the notion of the matrix of domination or intersecting vectors of power [26] to describe the way in which different groups, with different encounters with discipline and power and privilege, have only partial perspectives. Collins notably also posits intersectionality as a theory in the making—a “way of thinking” [28] and it is in that spirit that we adopt this lens and embrace the messiness of attempting to understand the extent to which experiences with a matrix of oppression and also what those matrices look like can sensitize HCI scholars and designers to non-normative ways of thinking about privacy and risk as it relates to technology. To do that, we borrow from the core tenants of Collins’ matrix of domination: interpersonal (how people's actions shape power relations), disciplinary (which rules apply, to whom, and when; e.g., bureaucratic organizations perform routine surveillance for the sake of efficiency), hegemonic or cultural (conditions under which power takes hold) and structural (how powerful institutions are organized; e.g., laws, policies, etc.) domains of power [26, 28, 29]. We map the construction of privacy imaginaries like the Handmaid's Tale on to hegemonic domains of power. Disciplinary domains and structural domains also map to tech company business logics and algorithms that fuel the accumulation of data for advertising. That is one of the ways that this matrix manifest is through surveillance capitalism/data colonialism, which reifies oppressions through advertising models, policies, and algorithmic surveillance that discriminates and disproportionately affects and harms certain marginalized groups. To grapple with this relational complexity, we looked to analytical approaches that have emerged from intersectional studies that focus on intercategorical thinking which requires that “scholars provisionally adopt existing analytical categories to document relationships of inequality among social groups and changing configurations of inequality along multiple and conflicting dimensions” [67].

We expected that because surveillance disproportionally impacts marginalized individuals (e.g., [18, 21, 38, 39, 41, 42, 84, 86]) potentially affecting their interaction and agency with technology that they would indeed be more likely to adopt non-technological approaches—feeling that they had no other options. But we were nevertheless not sure of how perceptions of technology and identity might interrelate, nor how they might affect participants, not in traditionally marginalized groups. Surveillance studies offer researchers a way of thinking about the collective nature of privacy and the way that power influences agency [65].

4.1 Laws, Reproductive Risk, and Privilege

Most participants have an idea of whether abortion is legal in their state, and this is often given as a reason they do not have to worry about their technology use. In a sense, the domains of power that form Collin's matrix are not in play (abortion is legal) so participants don't feel the need to take major steps to protect their privacy. Even for many of those who are in states that are hostile to abortion (e.g., where changes in elected leaders might usher in changes to their state's constitution in ways that criminalize abortion) or where abortion is currently illegal (or thought to be), the strategy usually doesn't involve technology e.g., to have a plan for moving or going across state lines for an abortion. Being in a location where abortion is not banned seems to confer privilege that overrides individuals’ race or ethnic backgrounds. Where we later see experiences overriding privilege is when reproductive risk level comes into play and/or some combination of illegality and experience with oppressive states or interpersonal relationships.

• Stephanie describes not being concerned because abortion is legal in her/their state but says that it would be worrisome if she/they lived in a state where it was illegal:

  I mean, I'm fortunate. I live in a state that still allows it. So, I'm not as concerned. I mean, if I were in maybe Alabama or Texas I would be more concerned. So, I think it depends on what state you're in and then what age you are.

• Stephanie who is in marketing in the technology sector describes having access to professional channels and friends who discuss technology strategies for reproductive privacy. Stephanie says that if she/they lived in a state where it was illegal to get an abortion she/they might take precautions like encrypted messaging and concealing her/their IP address to book appointments or travel related to reproductive health: “Yeah, I mean, I think if you're in a state that bans it you might have to use the incognito search on your Internet searches, maybe, so you can't be tracked if you try to register for appointments or maybe, if you need to book travel outside the State. Probably just have to protect your Internet privacy and your messaging, so probably need to use, um, Telegram or Signal for messaging services that don't track or keep a record of your conversations.”

  Clearly, Stephanie views location as a privilege to not engage with privacy precautions. Stephanie does not take these precautions her/themself because, as Stephanie points out, she/they live in a state that does not ban abortion. It is not that Stephanie doesn't know what to do, rather location privilege shapes the decision to not bother with these measures.

  With Victoria, we see where privilege, in the form of state law, is considered a reason to feel safe with technology, but that nevertheless it seemed wise to delete her/their period tracking app. Victoria, who is older with kids and not looking to get pregnant, was disappointed to delete it because she/they say it was very useful postpartum. Victoria says that even if she/they felt safe in her/their state, it is still a good idea to avoid using a period tracking app:

  Um! I had previously been using a period tracking app which was just really helpful, especially in the post-partum area when periods could be all over the place. I am feeling like, even though I'm in a state that's quote unquote safe right now, like I, I probably should avoid using that app, which is a bummer, because I really don't want to use a paper and pencil calendar um in terms of people.

• Victoria goes on to describe learning that there are data privacy issues associated with period tracking apps and that companies are not always transparent about what they do with the data. There were also a lot of people on social media saying to delete it and some news media stories about how large companies were sharing data with authorities:

  I did some research. It involved digital things I already knew. There are some data privacy issues, and not all companies are transparent about what they do with their data or who they give them to. Um, but also, I saw um just as a regular person on a lot of social media spaces people saying, ‘delete your apps, those are going to be tracked and shared.’ Um and it, you know you're starting to see some news stories, too, of large companies sharing some data with authorities.

• In essence, Victoria is willing to take action on things that seem obvious (like deleting a period tracking app) but isn't going to take any other steps in terms of digital privacy. In fact, Victoria was hesitant to go too far with researching digital privacy post Roe because she/they has a “hectic life” and didn't need to add to her/their worries.

Throughout our interviews, no matter what participants’ race/ethnicity, being in a state where abortion is not banned confers privilege such that participants did not feel that they had to take steps with their technology or otherwise conceal their behavior. When laws seem to become less clear or where there are elections that threaten abortion bans, we see a heightened sense of vulnerability. This vulnerability doesn't necessarily mean that participants will take the measures that Stephanie suggests, only that they might consider their options (e.g., what to do with embryos, whether to move to another state). When participants have additional added reproductive risk on top of experiences of risk with laws and culture, we see more dramatic steps being taken—making the argument for why this issue is intersectional. In the next section, we explore why technology intervention is not necessarily considered an option.

4.2 “The Handmaid's Tale” as an Outlook and Strategy of the Privileged

A number of participants framed concerns about privacy in terms of a Handmaid's Tale scenario in which the data on their phone, or collected about them by medical institutions, on the “dark web,” or by some “right-wing” activist organization could be used at any time to criminalize their behavior. Claims to Handmaid's Tale thinking are most prevalent among people who are older and feel less of a “target” for prosecutorial risk (i.e., do not feel the weight of cultural and structural domains of power) and may also have a low reproductive risk—not seeking to get pregnant or not likely to have a high-risk pregnancy (i.e., do not feel the weight associated with disciplinary and structural power)—possibly reading themselves into the dystopic storyline in which they are not vessels of reproduction because of their age. These women feel that literally and culturally they are not at high risk and in that way, eliminate another intersectional factor that might oppress them.⁹ This type of thinking makes sense because it gives them the opportunity to claim knowledge of privacy risks associated with the use of technology risks without allowing them to be entirely real—the Handmaid's Tale is, in the end, a fiction.

• Victoria is not too worried about the consequences of technology use because she/they is in a “safe” state, in a stage of life that is not a “high scrutiny bubble,” and because she/they is not actively seeking to get pregnant. That said, Victoria does consider what she/they describes as a Handmaid's Tale reality. For instance, Victoria wonders if she/they were to call her/their OB/GYN and ask to be seen immediately, would that raise red flags? For Victoria, The Handmaid's Tale is the sort of predictive algorithmic potential of the medical information she/they contributes to conversations with healthcare providers. Victoria does trust medical staff but notes that when leaving the hospital after having a second child, she/they wrote vasectomy on her/their record because they ask for a method of birth control. It's not that this information is incriminating, it's how much information gets out and possibly never deleted. Victoria imagines that in a Handmaid's Tale situation when data from five or more years is fair game it doesn't matter who you trust. The idea is that well intentioned individuals could document things that in a “Handmaid's Tale situation” could put Victoria at risk:

  Like, if there is a weird Handmaid's Tale situation, where, like, you know, data for the last five years is pulled, and they start making inferences based on certain data trends [recorded by healthcare workers].

• Victoria says that if this were two years ago, when she/they was at a higher reproductive risk and trying to get pregnant, she/they would be more worried about sharing information with medical professionals. Victoria didn't use IVF but considered it and wonders what would happen now if she/they used it. Victoria considers multiple scenarios like this that seem riskier in hindsight, working her/their way through risk assessments from then as if they were now.

  Victoria is a clear example of where heightened reproductive risk might influence digital privacy management, even in a state where there is legal “privilege.” Victoria's concerns also raise important questions about how well-intentioned care workers can essentially become the adversary. It also highlights how an understanding of how information is never forgotten (i.e., “data for the last years is pulled”) might make any attempts at future reproductive privacy seemingly futile.

  The Handmaid's Tale also came up for Elisa who did not feel she/they was a target living in a state where abortion was legal, in her/their 40s, and no longer seeking to get pregnant (i.e., low reproductive risk). When recollecting privacy technology choices in the days and months after Dobbs, Elisa recalls deciding not to sign a petition and would have deleted anything period-related on her/their phone, citing The Handmaid's Tale as justification for this thinking:

  So, like there was a petition to like fight against it. And I didn't sign that just because I was like. ‘I don't really want to put my name on anything because, I did watch the Handmaid's Tale.’ Still, you know it's fiction. So, I did not remove anything from my phone (but I did hear that people were) because I didn't have anything really on my phone … with regard to my period.

• Elisa decided not to engage in certain online activities that communicated her/their political stance and would have deleted a period tracking app if she/they had one, but was not planning on taking any other measures because of her/their perceived privilege (location, age, and reproductive risk).

  Emily, who is young and has had children, feels relatively safe in her/their current state where there is no trigger law and there is an effort to put abortion on the ballot for state constitutional amendment. Yet Emily's experience living in a more repressive state influences her/their plans for travel and what technology she/they will use when crossing state lines. That is, Emily is an example where experience with oppressive states influences perception of risk.

  Although Emily does not explicitly reference it, these are the types of things that are part of her/their Handmaid's Tale thinking: Emily deleted a tweet about the Dobbs decision, in retrospect because it seemed not in keeping with her/their account, which is professional, and also because of fears of ending up on some list of those being surveilled for future prosecutorial action. Emily also worries that her/their husband is from a state where abortion is not legal and where there are rumors of politicians or organizations (she/they are not sure) assembling lists of women who have had abortions. For this reason, Emily worries about crossing state lines and about GPS tracking. Emily also did call up her/their OB/GYN after Dobbs regarding her/their embryos—i.e., should abortion be outlawed, unwanted embryos might also be. Emily also worries that her/their medical records, even though they are supposed to be deidentified, are shared with the government for quality checks. Emily considers that online shopping could put someone at risk but is less worried about that because of her/their modest online shopping habits. Despite Emily's assertion that her/their online shopping habits are modest enough not to be a concern, she/they acknowledges that virtually any shopping online or with credit cards linked to reproductive health could result in someone learning they were pregnant.

  Oh, gosh! I mean, I guess like things like Amazon, I mean if you have like an Amazon subscription for like pads or tampons, and then you stop it, I guess, or even like, you know, I've heard some of my friends when they got pregnant like, somehow, they will get like formula sent to their house, or something like [that]. I haven't had that happen to me personally. Um, but I've heard of that happening, so I guess credit cards, or, like Amazon subscriptions, would be like another thing … I'm not someone that like has a bunch of subscription services, or like, had a subscriptions to like daily products.

• While Emily's location confers privilege, the experience of growing up in, and traveling to (currently) repressive states compounded with reproductive risk make her/them more cautious traveling to another state, where GPS tracking might be a concern. Emily is aware of the structural risks because of experience with cultural or interpersonal power.

  Lily, who is mixed race and prefers to self-describe her/their sexual orientation has had experience with abortion under circumstances following rape and other accompanying trauma. Lily still mourns the child she/they lost but also knows it was the right decision for her/their education and career. Lily is an activist for human rights and, like several of our participants, has a passion for reproductive education. At 40, Lily is trying to conceive and facing new medical complications that make an ectopic pregnancy an increased possibility. Lily decided not to download a period tracking app after hearing on NPR that it was not safe. Lily also heard that IP tracking was a concern and would consider that if she/they had to look for abortion clinic. Otherwise, Lily hasn't considered anything else saying that we live in a Handmaid's Tale reality (which she/they equates with Gestapo and Nazi Germany) currently and she/they essentially feels helpless:

  No, it's just. I was like. Are we really gonna get to that point where it's kind of like the Gestapo like in Nazi, Germany like where you can … I never watched the Handmaid's Tale to be honest because … It actually could happen. And that's why I've never watched it because it's real life … I know what the gist of the show is, just because everybody posts about it. But it's almost like we're already there like that. They're going to be able to just access. And if they pass these laws that you know where they can criminalize you, they're going to have the FBI or CIA anybody investigating. You can have your search results from your computer.

• Like the other participants in this category, Lily envisions a Handmaids Tale reality in which lists are being created and IP addresses are being tracked. Lily also worries for the younger generation who are of a more fertile age and might not be ready. Lily speaks of “generational trauma” and that her/their mother didn't want to be a mom and about abuse in her/their household before saying, “It's cruel to not have the choice, because it's a decision between you and your doctor.” But for Lily, perceptions of risk are narrowly constructed around changes to the legal climate and medical circumstances. Lily has several factors (location, reproductive risk, and past experiences with trauma) that might make her/them more cautious. But Lily's outlook is shaped by a sense of helplessness (“They're always tracking us”) and the knowledge that the laws in her/their State do allow for abortions in certain medical circumstances that would apply. Although among the strictest in the country, her state does make exceptions for the life of the mother (and in cases of incest and rape), which she/they finds reassuring.

Not all participants explicitly talk about a Handmaid's Tale but maintain a nihilistic view that they are always being tracked and that they can't do anything about it and/or don't want to live their life worrying about it constantly.

• Kim who is trying to get pregnant with in vitro fertilization (IVF) and has, in fact, increased her/their use of social media to keep tabs on the laws regarding embryos says that, “I feel like Facebook and Google are always listening to us, and I feel like everything we are discussing because I have the Google Assistant in my house and every room, and I feel like you know anything that we discuss in the house or share online or communicate online. All these different corporations are collecting all that information, and I don't know how they're being used … I mean there's only so much I can do, because you can only protect yourself so much. I can't be like really, picky, so I mean right now it just. I can [accept being tracked].” While Kim is privileged in terms of her/their location she does have reproductive risks. Kim seems to be aware of her/their resignation and it seems to play a big role in her/their lack of concern for privacy management strategies.

• Veronica provides a case of someone whose higher reproductive risk does lead to concerns about privacy but a sense of helplessness or apathy in terms of technology surveillance results in her/their not taking any digital privacy management strategies. Veronica is seeking to get pregnant and is at a higher risk of miscarriage. Veronica does worry about the repercussions of sharing her/their fertility journey with people because she/they doesn't want to be judged. But knowledge of this cultural or interpersonal power does not extend to concerns about prosecutorial repercussions despite some concern about laws. When it comes to technology, Veronica says that even though she/they had friends who worry, it isn't going to change anything: “To tell you the truth, you know I don't know, but like. They are for sure for sure. So, some things they just they don't really like to talk around their phones. They turn them, like, if they having deep conversations, they will turn their phones off … But I kind of, I don't want to live like that, you know?” Veronica is not positive that there is near total surveillance of her/their phones as friends suggest but is also not willing to make drastic changes in order to guard against it.

4.3 Young (Cultural Domains of Power) /High Reproductive Risk (Disciplinary/ Structural Domains of Power) and Practicing No/Low-tech Strategies of the Less Privileged

Those who are younger and sexually active tend to be preoccupied with no/low-tech technology privacy strategies like good contraception and being choosier about partners, and for these reasons feel safe. Being younger equates with higher suspicion (cultural domains of power) and higher reproductive risk (disciplinary and structural domains of power)—not only because they may be more likely to get pregnant but because several cannot or choose not to be on birth control. There are some younger participants for whom risk of high-risk pregnancy outcomes also impose disciplinary and structural power because abortions (if there are complications, genetic issues, etc.) or embryo disposal (i.e., for those who struggle to get pregnant) may be outlawed.

A few participants like Rowan stressed being attuned to their bodies as a way of dealing with these power differentials:

• Rowan is a bisexual, multi-racial woman who is very concerned with reproductive health education. Rowan says that abstinence and safe sex are the first defense against changing laws about abortion. Rowan says that she/they has been feeling this way for a while, but that Dobbs “solidified” that resolve to be more conscientious about sex—choosy about partners and more aware of her/their body. In the event that Rowan did get pregnant (and Rowan really expects that will not happen), she/they would go to a planned parenthood.

  To be honest, I just. I feel like I'm in the position where I just rather … be as cautious as possible to not have you put in that predicament. But if matters do arise. Well, I know what to do with that situation.

• Rowan did delete her/their period tracking app but is primarily focused on how to prevent unwanted pregnancy with her/their philosophy of choosing partners and having conversations with them upfront—which Rowan attributes, in part, to her/their race and sexual identity. Roman also has the added concern of not wanting to go on medication birth control, so conversations about other forms of birth control are necessarily consensual.

  I'm bisexual so like that's kind of like from the get-go. So, I kind of can like gauge … if they don't like LGBT people, they're probably …. it's further than just like talking or like engaging with someone like from like the get-go. But I would say, um, in terms of the conversation about like reproductive health. Kind of talking about, ‘Hey, like, I want to use condoms, and I do not want to get pregnant. I don't want kids. I'm sure you don't want kids right now, at least not with me.’ And kind of just be able to have this conversation, because the fact of the matter is, I'm not going to go on birth control, because I don't want to, for my personal reasons. Um. So you're going to use a condom if you're going to have sex, and if not, then I guess you won't be having sex and like that's really as simple as it gets, because I'm not going to compromise that me personally.

• Rowan's age (cultural power) as well as reproductive risk (not being on birth control) and to a lesser extent race/identity shape her/their reproductive privacy strategy. Rowan manages these cultural, disciplinary, and structural domains of power with abstinence, better cycle management, and partner selection—i.e., non-tech strategies.

Other young participants acknowledge the risks of reproductive data while at the same time, pointing out that people should also just be careful about their sexual practices. For instance, Antonia considers that browsing histories and online tracking might be a concern but prioritizes what she/they can control, which is her/their body—and tends to adopt no/low-tech strategies. Antonia also has had some bad experiences with healthcare facilities (including in the context of abortion) and would rather avoid trusting anyone when it comes to advice about reproductive privacy decisions. Antonia recounts a harrowing experience with healthcare workers during an abortion in which they seemed unresponsive to her/their medical needs and unwilling to provide even a minimum of information including the side effects of abortion medication.

• I don't know if I should be careful about browsing and incognito tabs and being worried about like, my data tracking, because I know it's getting crazy … We really need to be in control of our own bodies, and that's just me, I guess I'm untrusting, because I've never had a doctor who I've had a trusting relationship with ever so it's been me on my own.

  Antonia, like a few of those we interviewed, is not familiar with the law (disciplinary and structural power) in her/their current state but would figure it out should something happen. Antonia feels that being from New York (where her/their family is based) means that she/they is safe because it is still legal (“for now”). Interestingly, Antonia perceives that she/they is “privileged” by her/their location, even while her/their inability to be on birth control puts her/them at even higher reproductive risk.

  Okay. So me, I'm privileged. I'm lucky I'm from New York. So if I had to go back home, or if I had to get something to delivered somewhere, I'm safe in New York for now, for now that's just for now.

• That said, as a bisexual individual, Antonia describes being conversant in the “coded” language that she/they and other marginalized friends have used to keep themselves out of “Facebook Jail”—a timeout that Antonia has experienced when Facebook suspends accounts for violating community standards. Antonia talks about how experience (particularly with friends that are BIPOC) have given her/them insight into and practice with coded language (e.g., using “wheat” for “white,” or a symbol of wheat so you don't get flagged for saying “white nationalists” and so on, terms that Facebook has banned). Antonia believes she/they knows well how to use “coding” to get around algorithm/surveillance and thus navigate an abortion through online channels if necessary. When asked, Antonia mentions coded language used to locate people who will help with abortions in other countries.

  I also have a large community of people online who kind of have, like an underground railroad system going, if you need to come somewhere if you need [code word for abortion]

• Antonia, who is young and not on birth control and does not trust healthcare institutions, manages disciplinary and structural as well as cultural and interpersonal domains of power through reproductive education, off-line cycle regulation, and a digital strategy involving low-tech coded communication—a kind of subversive cultural power. In a certain sense, Antonia represents the kind of no/low tech “literacy” that is required to navigate a multi-vectored and highly contextual set of risks.

Several of the younger people whom we spoke with cannot be on birth control for health reasons.

• One in particular, Rocky, feels quite frightened. In fact, Rocky and her/their husband moved from a country where abortion is illegal to the US in the hope that it would be better for them. This person discussed considering whether to relocate to another state because theirs was currently hostile to abortion rights. Rocky has always been cautious about using Google but is now more so since the Dobbs decision. At the same time, Rocky perceives that she/they will be safer if she/they moves to a state where the right to abortion is guaranteed by law saying that “I will be more comfortable at least sharing my decisions online, or getting back to using Flow [the period tracking app]. Because I know whatever happens law would support me.”

  Rocky has a high reproductive risk because of the fact that she/they is young (cultural domain of power) and cannot be on birth control (disciplinary and structural domains of power), has experience living in a repressive country (cultural and interpersonal domains of power), and is in a location where abortion is under threat (structural domain of power). Rocky considers mostly non-technical solutions—i.e., moving to another state, not using Google search. Rocky also manages power differentials with the prospect of regaining structural power by moving to other states.

4.4 Reproductive Risk (Disciplinary and Structural)/Experience Risk (Cultural and Interpersonal Domains of Power) and More Tech Savvy Strategies of the Less Privileged

We found that participants who have experienced disciplinary/structural or cultural/ interpersonal risk were more prone to take steps to secure their privacy and to do so in a way that combined high and low-tech approaches. Sylvia is young (cultural domain of power), has experience with repressive environments (cultural and interpersonal domains of power)—including living in a state that bans all abortion (structural domain of power) and has seen how pro-choice stance can threaten job opportunities. Jessie is pregnant, in a state that bans abortion, and has high reproductive risk (disciplinary and structural domains of power). Both employ more sophisticated technical strategies for the sake of their reproductive privacy, Sylvia primarily because of past experiences with social/job risk (cultural and interpersonal domains of power) and Jessie because of her/their reproductive risk (disciplinary and structural domains of power). Neither one feels they are targets per se but worry more about the complex nature of surveillance offline and online.

• Sylvia worries about the repercussions of sharing information with people in her/their personal and work life who might not agree (cultural and interpersonal domains of power). She/they have experiences with discrimination in the past and worries her/their views could have negative implications on her/their career. Sylvia doesn't think she/they is on the “radar” of the government or pro-life activist “vigilante group” necessarily, but considers that there could be implications from her/their data (private conversations on social media, search history, etc.) being used in nefarious ways by people seeking to prosecute or even protest outside clinics. Sylvia explains that the data used by marketers to profile consumers could also be used to target someone and then hack into their personal data. Sylvia has also had friends who have written privately about their pregnancy complications and then received hate mail; in her/their view the privacy of social networks has “been compromised.” Sylvia makes connections between online behavior and violence friends have experienced outside clinics, noting that she/they doesn't think online behavior was the vector but that it could be. In Sylvia's view, the biggest concern would be if somehow her/their online data were used for personal attacks, which Sylvia doesn't believe is that likely, given that she/they is not a public figure:

  I guess there's a couple of possible scenarios. One is like a larger company, say Google or Facebook. These companies that have access to so much data, using my data points among the conglomerate of millions of other people for me. That's less of a concern. The bigger one would be like a personal attack. And realistically, I don't think that I'm kind of visible in the public sphere enough to be at risk for this, or to be at a large risk … whether someone's hacking your computer, whether they're installing some sort of monitoring software. I think that the effort required to do that … would be much more likely to be applied to someone like a public figure, a politician. I don't know a well-known actor or actress whatever, so. I don't see myself as the main target of this. But I'm just, as I say, I want to take every precaution possible, especially since, in my case, the cost of taking a precaution is relatively minimal; and though I don't see this as a directly pertinent risk, I think the consequences should it happen should someone be tracking me? Um could potentially be very severe, and I would love to avoid that, if possible.

• Sylvia is primarily worried about people finding out about her/their abortion stance. The precautions that Sylvia is taking, to essentially avoid cultural and interpersonal domains of power, include low and high-tech strategies like being more careful about what she/they says online and using a VPN. This “minimal” effort Sylvia believes is worthwhile given the severe consequences, which to her/them are vivid: “In this case. I see this more as a hypothetical, but I definitely think that sort of big brother scenario, be it a government agency, or some sort of vigilante group in terms of monitoring individual data is a possibility. And so again, I see this as more a bit more removed from my particular scenario. But um! That scares me and um I've heard of just so many stories about people going to clinics and being, I even know someone who went and was like followed … like some larger group or organization tracking my day.”

  Jessie, who is currently pregnant and queer, was aware that her/their medical condition put her/their at an increased risk for miscarriage (disciplinary and structural domains of power) or the need for abortion. In Jessie's case, she/they is in a state where abortions, even for ectopic pregnancies, are illegal (structural domain of power) and so had a plan for how to get it done in another state. The plan involved VPNs, cash, and other legal considerations. Jessie is thus cautious, but also felt that she/they wasn't the target of investigations because she/they felt that poor women who do not have people to advocate for them are the primary targets: “The women that have been prosecuted for miscarriages and things like that have been poor and unable to advocate for themselves. They are easy targets. The worst that happens is, we have to leave the state.” Jessie is taking precautions with VPN, with browsing, but does not anticipate prosecution. At worst, Jessie would have to leave the state.

There was, however, one participant for whom location alone (structural domains of power) was a motivator for low- and high-tech strategies. Lucy's perception that abortion is not legal (though it is in her/their state) has made Lucy more careful about sharing online. Even before the Dobbs decision Lucy was careful about use of social media. Lucy turned off Google tracking whenever she/they “remembered” and wasn't using it for navigation. Lucy is also careful about passwords. Lucy says that she/they take precautions because of “bad stories” she/they have heard about friends being scammed. If Lucy were to have an unwanted pregnancy it would mean going to another country. Lucy also mentions that, while very careful, the overturn of Roe v. Wade might compel her/them to use double contraception.

5.1 Using an Intersectional Interpretative Lens in HCI

In our results we describe how certain overlapping categories (age/reproductive risk, location/laws, and experience with oppressive states or environments) shape privacy strategies and the ways those map to Collin's matrix of power. Living in states where there is no abortion ban and having low reproductive risk (i.e., lack of experience with powerful domains) equates with a Handmaid's Tale way of thinking about privacy management which is, we argue, privileged. By contrast, being young and/or of higher reproductive risk (fertility issues, sexual activity, inability to use contraception, risk for miscarriage, ectopic pregnancy or genetic conditions, etc.), living in states that ban abortion, and/or experience with oppressive states or environments (trauma, distrust of medical professionals, workplace discrimination, etc.) all fall along one or several axis on Collin's matrix and lead to no/low-technical or more technical strategies, or some combination. It is evident that these strategies may equate with type of risk. For example, younger participants seemed to be more invested in old fashion period tracking, selective partnering, and abstinence (all no-tech strategies) than their older counterparts.

Those who did not live in a state where abortion was banned and who were not at high reproductive risk had an outlook consistent with what Collins conceives of as hegemonic or cultural power under which matrix of domination takes hold. These participants perceived that we live in a kind of Handmaid's Tale world where reproductive freedoms are no longer guaranteed. But, for now, they depend on Collins’ structural power (how laws and policies are implemented) to maintain their status as “safe.” Future research should explore the extent to which concern about structural power (e.g., law enforcement) translates to technology behavior. Perhaps to better understand privacy strategies and perceptions, HCI researchers should explore these metaphors about power through elicitation.

By contrast, those who were younger and oppressed by cultural power were more likely to rely on privacy strategies that avoided technology altogether, perhaps because of their perceived visibility as being of reproductive age. Fear of cultural power may be an important construct determining privacy strategy, and one that is related more to social surveillance, where the vectors of threat are more ambiguous and perhaps, more low tech. Those who perceived themselves to be at high reproductive risk (disciplinary and structural domains of power) or living in a state that banned abortion (structural domain of power) were more likely to take a combined high and low-tech approach, perhaps because of their privilege and perceived lack of visibility, which, for example, Sylvia emphasized. They also may have felt their threat had more technical advantage. HCI researchers should be attuned to these complex and shifting power matrices that influence changes to perceptions, use, and receptivity to privacy strategies.

By framing our study around intersectional relationships of power, we were able to look at a less typical interrelated axis of power involving reproductive risk, location/laws, culture and personal experience. Finally, HCI researchers should approach studies of technology use and privacy risk by first exploring the way that identity facets and intersectional characteristics influence perception of risk and mitigation strategies with technology, and, in particular, be attuned to strategies that do not involve technology.

5.2 Examining the Media and Argument for Using a More Critical Lens in HCI

It should give privacy researchers pause that media emphasis on the risks of period tracking apps was apparently so effective that all of our participants erased them from their phones. As noted, period tracking apps, while notoriously insecure, are essentially no less harmful from a tracking and data triangulation standpoint than other apps people download to their phone. While period tracking apps are selling period data, many other pieces of data can be used to detect reproductive health status, including search and location data and purchases (e.g., menstrual pads). Perhaps providing insight into these practices and basic guidance about obfuscation [22] could go a long way. HCI researchers should be critical of how technology privacy is portrayed in the media and its potential for shaping norms that they in turn study (e.g., [69]).

It may be too late, as some participants acknowledged, to take back medical health histories that put them on “abortion lists,” but understanding (or being reminded) of what medical histories reveal could support small changes to behavior. Obviously, there are tradeoffs between providing accurate personal health data (for example, like contraceptive methods as one participant pointed out) and protecting one's privacy.

Overriding this is the media's lack of attention to privacy's intersectional risks. Our study only begins to scratch the surface, but it is clear that when it comes to privacy no one size fits all. Individuals need help sifting through what are reasonable reactions to legitimate threats, and those differ depending on the person and circumstances. The response from the media was an urgent message encouraging all women to delete their pregnancy app, followed by a set of articles saying this was not nearly enough. A better approach might have been to provide information about vectors like application location tracking, browsing history, unencrypted messaging and so on and then provide adversarial strategies like reporting to authorities that result in warrants, organizations that pay for third-party data, and others so that people could decide where their vulnerabilities are and what to do about it—be it turn of their location tracking, leave their phone at home, or be careful about who they talk to. In some sense, consideration for different domains of power would be helpful. For instance, consideration of vectors of threats from health care workers that are suspicious of miscarriage (disciplinary domains of power) or being wary because of currently or in the past living in environment in which abortion is looked down upon (cultural domains of power) that could leave someone more susceptible to reporting or simply an understanding of the laws and what they ban and when (structural domains of power). These are no fail-safe measures but we should inform the public in a way that helps them think intersectionally and contextually about their risk, and thus allow them to manage it.

Based on our research alone, the media should focus on guidance that is geared towards legal climates and reproductive risk. The media could frame information that would support privacy decisions around state laws and reproductive risks. For instance, individuals should be aware that IVF puts them at an increased risk for ectopic pregnancy, which puts the pregnant person at risk for death and some states do not have exceptions for terminating ectopic pregnancies. This was a deciding factor for Lily. HCI researchers need to be attuned to media narratives and missteps and consider how that might influence their research design, or provide avenues for critical research.

5.3 The Implications for HCI Privacy Thinking and Framing

5.4 Guidance for HCI Studies of Privacy: Through a Privacy Intermediary Lens

The issue for studies in HCI and related privacy fields is then, whether and how we begin to study privacy in terms of the things that people do that do not always involve technology and how they perceive power structures, and their place in the world of consequences.

While, as HCI privacy researchers, we are not privacy intermediaries—that is, we are not those providing health, counseling, and other services to people who may become pregnant—this orientation can be used as a lens to think about how we do research, and also how we do so in a way that supports privacy intermediaries. That is, privacy intermediaries is both a concept and frame—a way of thinking about and orientation toward privacy research—and a role performed by those who provide privacy management guidance.

Below is our guidance on how privacy scholars might conduct research. These guidelines are drawn from findings from this research as well as the concept and role of privacy intermediation [71]. While we seek specifically to include privacy intermediaries in research, we also intend for this set of guidelines to move researchers to think differently about their research:

• First, we urge researchers to identify and include privacy intermediaries in research. These could be service providers that people depend on for guidance, in this instance, health care workers, social workers, and so on. Rationale: McDonald et al. [71] has highlighted the role that privacy intermediaries can play in development of privacy strategies for marginalized communities and for helping researchers to think about the intersectional facets of those they seek to study but which they cannot necessarily access. McDonald et al. and our research demonstrates the critical importance of discovery of these facets and the need to include those with sightline into these complex facets in research about populations at risk. Research with privacy intermediaries has the potential enhance the following guidance:

• Guidance: Researchers should have an appreciation for the risks and strategies employed by individuals whose surveillance has consequences not readily experienced by the mainstream. Rationale: We have shown that surveillance risk is important to understanding privacy strategies. Historically marginalized individuals provide insight into surveillance risks because they experience them more (e.g., [18, 21, 38, 39, 41, 42, 84, 86]) and when they do experience them the consequences are graver.

• Guidance: Be willing to contend with fluid categories and think about privacy risk intersectionally, as a stack that is person-dependent. Rationale: We found that, much like Collins’ work [28] advocates, this issue of privacy management was highly contextual and based on categories of risk that were not strictly race/ethnicity based (a common understanding of intersectionality) but on other dimensions of identity and experience that are somewhat more difficult to map but are still important [29, 33]. (See more on the role of race/ethnicity in the next section).

• Guidance: Think of “technology literacy” as something that both requires other “literacy” and is so context-dependent that it is perhaps not useful as a framework for thinking about privacy management (including identifying its deficits). Rationale: For example, as we found, knowledge about reproductive risk was perhaps more important than understanding technological risk. Moreover, and somewhat relatedly, some of the strategies employed by participants were not (or seemingly not) technologically advanced like managing one's cycle, using coded language (echoing previous research [14, 34]), or using birth control, suggesting that traditional notions of “technology literacy” are simply not useful. In some cases, participants felt that reproductive knowledge and offline behavior was more critical to securing their privacy, suggesting that concepts of “literacy” need to be profoundly broadened, if not rethought. Even if we don't abandon “literacy” as a concept, we still need to explore how to provide people with feasible strategies and a level of privacy “literacy” that goes beyond a tool-centered education. We need to consider the many facets of privacy protection that are necessary to achieve a more holistic proficiency over reproductive privacy, given the numerous and highly varied threat vectors to which menstruating people are exposed.

• Guidance: Embrace no/low-tech strategies as legitimate (maybe sometimes, the only) option. Rationale: We learned from younger participants, in particular, that no/low-tech strategies (e.g., like partner discretion and condoms) felt like the best option for them. This echoes previous research that has shown that no/low technology strategies and digital abstinence are sometimes adopted by “at risk” and marginalized individuals who might avoid digital record-keeping, rather than look for ways to safeguard their digital records or engage in digital abstinence [43, 71, 98, 101]. These strategies have been discovered to include “privacy intermediaries” who offer unique sightline into intersectional challenges face by marginalized communities [71]. For HCI researchers to embrace no/low-tech strategies, they must consider “literacy” and “agency” beyond technology, and conceive of users in a more holistic and intersectional sense. For example, does doing this type of research require a deeper understanding of people's reproductive health knowledge—as Kumar et al. discover [63]?

• Guidance: Model good privacy behavior. Rationale: It was really important that we were strict about participants obscuring their names and photos in video and having a limited conversation with us online, and so on. These behaviors are important to normalize when doing research. It also keeps researchers from being vectors of privacy risk, and forces us to think about how we could ourselves be adversaries, guidance that is drawn from study of privacy intermediaries in other settings where they are well aware of their potential role as vectors of threat [71].

• Guidance: Explore metaphors that could translate/bridge to conceptualizing more favorable policy concepts like those outlined by the EFF [72]. Rationale: Many participants offered up The Handmaid's tale as a way of thinking about surveillance capitalism and authoritarianism. Indeed, we had participants summing up vast surveillance infrastructure with examples like, “you know when you see sweatshirt ads.” Future research might explore how these concerns line up with policies that are being recommended in government or by advocacy organizations.

• Guidance: Be wary of assuming that, as privacy researchers and experts, we have any perspective on the “right” way. Rationale: We came into this research thinking or hoping that participants who were at risk for unwanted pregnancies would say that they were using specific privacy technologies and had to accept that better use of birth control was a reasonable low-tech solution.