Trans Workers and the Real Conditions of Online Jobs and Internet Work

Eight months into a retainer with a mid-size SaaS company, all remote, all Slack and Figma and Tuesday standups, they moved me from a 1099 to a W-2 because finance wanted the headcount cleaner. Routine. HR sent a link to a screening portal — Sterling, I think, though it may have been HireRight — and the form asked for "other names used." I typed my legal name, which is not the name anyone at that company had ever seen. The report came back three days later. A week after that, the offer got restructured into something part-time and worse, and nobody said the word "trans" at any point in the conversation.

I have no proof. That's the whole shape of the problem. There is no version of that story where I walk into a hearing with a document. I worked as adult Cam Site Transgender like I explained on YOOCO and adult Cam Industry for a Trans is a slur fiesta even more in porn where you are treated just as a fetish Tag for "Trannies Lovers"!

I want to write about what online jobs and internet work actually look like for trans people right now — in 2026, after a year of specific and dateable changes to identity documents, federal enforcement posture, and the platform economy that a lot of well-meaning "remote work is a refuge" articles were written before. The refuge framing isn't wrong. It's just about fifteen percent of the picture, and the other eighty-five is administrative.

Why the channel matters at all

Start with the push, because it's well documented and the numbers are grim in a way that's easy to skim past.

The Williams Institute at UCLA Law surveyed LGBTQ adults in the workforce in summer 2023 and published in November 2024. Among transgender respondents, 82% reported experiencing workplace discrimination or harassment at some point — fired, not hired, not promoted, or verbally, physically, or sexually harassed — because of sexual orientation or gender identity. In the past year alone, 47%, against 15% for cisgender LGBQ employees and 29% for nonbinary employees. Two-thirds, 67%, had left a job over how they were treated, roughly double the cisgender LGBQ rate.

The mitigation behaviors are the part I'd underline. Seventy-one percent reported covering: 39% changed where, when, or how often they used a bathroom; 46% altered their voice or mannerisms. Thirty-six percent were not out to their current supervisor.

One honest caveat that the coverage of that study almost never included: the transgender subsample was 86 people, drawn from a total sample of 1,902. That's a small n for the precision those percentages imply. The findings are directionally consistent with everything else in the literature, which is why I'm citing them, but if you see "82% of trans workers" quoted as though it came off a Census file, it didn't.

The larger instrument is the 2022 U.S. Transgender Survey. Unemployment among respondents ran 18%, against roughly 3.7% in the general population that year. Eleven percent of respondents who had ever held a job said they'd been fired, forced to resign, laid off, or lost the job specifically because of gender identity or expression. Thirty percent had experienced homelessness at some point. A 2024 analysis using federal administrative data including tax records found a significant wage gap versus cisgender workers across several specifications, which matters because it's one of the few findings here not resting on self-selected survey samples.

Sixty percent of trans people in the Williams workforce sample were earning under $50,000. Half were people of color. Seventy-one percent were under 35.

Against that, remote work offers one specific thing: control over the timing and terms of disclosure. Not privacy exactly — control. You decide whether a client learns anything about your body or your history, and when, and the decision isn't made for you by a hallway.

Researchers found this early. A 2019 position paper on trans software developers identified the same three themes across interviews: identity disclosure control, high-impact technical work that speaks for itself, and the autonomy to disengage and re-engage on your own schedule. Laura Doering and András Tilcsik's 2025 study in Organization Science, surveying 1,091 professional women working the same job both remotely and on-site, found consistently less everyday gender discrimination in the remote condition, with the effect strongest for younger women and those working mainly with men. Different population, same mechanism: fewer unstructured in-person moments means fewer opportunities for the thing to happen.

That's real and I don't want to undersell it. But.

Harassment migrates; it doesn't evaporate

Project Include surveyed 2,796 tech workers, 97% of them remote. Thirty percent reported increased gender-based hostility since the start of the pandemic. Among transgender respondents specifically, 42% reported increased gender harassment.

The tooling is a real part of this. Slack, Zoom, Teams, and every project-management surface people spend their working lives in were not built with harassment reporting as a primary function. There's no equivalent of walking to HR. There's a DM thread and a screenshot, and both parties know that the screenshot proves the message and not the pattern. Project Include's respondents also reported not escalating — sometimes because the hostility was coming from HR or leadership, sometimes because they didn't believe anything would follow.

And a hostile remote environment has a specific cruelty to it: it arrives in your home, on the device you also use for everything else, and there is nowhere to go afterward that isn't also where it happened.

So: remote work removes a category of exposure and adds a different one. That's the accurate framing. Anyone selling it as an unqualified win is selling.

The part nobody writes about: online work runs on identity verification

Here's the thesis I actually want to defend, because I think it's under-discussed and I think it's now the dominant practical constraint.

Physical workplaces gatekeep on presence. Online work gatekeeps on documents.

Trace the funnel for any remote role or platform gig and count the checkpoints. Application, possibly with an automated screen. Video interview, sometimes one-way and recorded. Background check through Checkr, Sterling, HireRight, or a smaller vendor, with an "other names used" field. Identity verification through Persona, Onfido, Jumio, Veriff, Shufti, or one of a dozen competitors, typically requiring a government photo ID plus a live selfie. Payment onboarding through Stripe Connect, PayPal, Wise, Payoneer, or Deel, each running its own KYC against AML obligations. Tax onboarding: W-9 with legal name and TIN, matched against IRS and SSA records. Then the ongoing layer — SSO directories, HRIS records, invoicing systems, and in gig platforms specifically, biometric re-verification at session start.

Every one of those is a place where a legal name or a sex marker can surface, and none of them was designed with any thought about who that surfacing costs.

An in-office job has one or two of these checkpoints, usually front-loaded, usually handled by a person you can talk to. Distributed work has eight or ten, all automated, all with escalation paths that route to a reviewer in another country who has your ID on screen and a queue to clear.

I know a translator who has done exceptional work for six years and who has, by her count, been locked out of four payment platforms for "document mismatch." Each unlock took between eleven days and four months. Nobody was hostile. The system just did what it does.

What changed in the documents, precisely

This is where 2025 and 2026 stop being background and become the operative fact.

On 20 January 2025, Executive Order 14168 was issued, published at 90 Fed. Reg. 8615, directing federal agencies to recognize two sexes and to enforce sex-based rules on that basis. Two days later, on 22 January 2025, the State Department overhauled the rules for sex markers on passports, reverting to pre-1992 practice: new passports reflect sex assigned at birth, no sex-marker changes, and the X designation — available since 2021 — discontinued.

The ACLU sued on behalf of seven transgender and nonbinary plaintiffs. In Orr v. Trump, a federal court in Massachusetts issued a preliminary injunction on 17 June 2025 blocking the policy, later certifying a class and expanding the injunction's scope. The First Circuit unanimously declined to stay it. The administration went to the Supreme Court, and on 6 November 2025 the Court granted the stay, allowing enforcement while litigation continues. Justice Jackson dissented, joined by Justices Sotomayor and Kagan. Solicitor General D. John Sauer had argued the policy was lawful and that the Constitution doesn't bar the government from defining sex in terms of biological classification. The case remained active into 2026 with cross-motions for summary judgment.

The practical translation, which Lambda Legal and HRC have both published in near-identical language: a passport you already hold with an accurate marker, including an X, stays valid until it expires. You don't have to surrender it. But any new application, renewal, or replacement will most likely come back with sex assigned at birth.

Sit with what that means for someone whose income comes through the internet.

Your passport expiration date is now a career event. It is the date on which your primary internationally-recognized identity document stops matching your presentation, your Stripe account, your client-facing name, and possibly your driver's license — and every KYC re-verification, every platform that decides to refresh its records, every new marketplace you'd otherwise sign up for becomes a disclosure event you didn't schedule. A lost wallet becomes a professional emergency. A replacement is not a replacement.

I have watched people reorganize their entire five-year plan around a date printed on a small book, and I have never once seen that dynamic mentioned in a remote-work guide.

The machines have a documented failure mode

There's a second layer here worth being precise about, because it's easy to overstate and the overstatement gets used to dismiss the real finding.

Automatic gender recognition — software that infers gender from a face — has a measured, published accuracy gap. Morgan Klaus Scheuerman, Jacob Paul, and Jed Brubaker published "How Computers See Gender" in the Proceedings of the ACM on Human-Computer Interaction in November 2019, testing commercial facial analysis services from Amazon, Clarifai, IBM, and Microsoft against a dataset of self-labeled Instagram images. True positive rates averaged 98.3% for #woman and 97.6% for #man. For #transwoman, 87.3%. For #transman, 70.5% — with Amazon's service at 61.7%. Non-binary genders were not classified correctly by any system tested, because the systems have no such category.

That sits alongside Joy Buolamwini and Timnit Gebru's Gender Shades work, which found commercial classifiers erring on darker-skinned women at rates up to 34.7% against 0.8% for lighter-skinned men. The failures compound. Os Keyes's 2018 survey found 94.8% of the academic AGR literature treated gender as binary in the first place.

Now, the precision I promised: most modern identity verification is not gender classification. It's 1:1 face match — comparing a live selfie against the photo on the submitted document — plus liveness detection to defeat printed photos and replays. That's a different task and it doesn't inherently misgender anyone.

The problem is what surrounds it. A 1:1 match against a document photo taken before or during transition is a harder match than the same check on a cisgender applicant, and harder matches route to manual review. Manual review means a human contractor comparing your face to a photo of a person who no longer exists, with a "does this look like the same person" prompt and a throughput target. Some platforms layer demographic inference on top for fraud scoring. Some run 1:N face search across the whole enrolled population to catch account renting, which is a reasonable anti-fraud measure that also means your face is a persistent index key across every platform using that vendor.

The gig-economy IDV market is competing on continuous re-verification — biometric checks at shift start, at app unlock, at first job accepted — and pitching "reusable KYC" that ports a verified credential between platforms. Read that as a trans worker and it means one bad enrollment can follow you.

The 1099 gap is the biggest legal fact in this article

Most people, including most lawyers who don't practice employment law, believe that Bostock v. Clayton County settled this. In June 2020 the Supreme Court held 6-3 that Title VII's prohibition on sex discrimination covers discrimination based on sexual orientation and gender identity. An employer who fires someone for being transgender violates federal law.

Two limits that matter enormously for internet work.

Title VII covers employees, not independent contractors. Freelancers, gig workers, consultants, 1099 filers, business owners contracting with companies — outside the statute. Not partially covered, not covered with a weaker standard. Outside. Section 1981, the Reconstruction-era statute sometimes used to fill this gap for contractors, reaches race, not sex or gender identity.

Title VII applies only to employers with 15 or more employees. A four-person startup contracting a designer is doubly outside.

So the worker most likely to be doing online work — 1099, multiple clients, no single employer — has the least federal protection available. And that's before you get to enforcement.

State law is the actual battleground, and it's uneven in ways that reward deliberate choices. California's Fair Employment and Housing Act, at Government Code section 12940(j), makes it unlawful to engage in discriminatory harassment against "a person providing services pursuant to a contract" — one of the few statutes that says the quiet part out loud. California's Unruh Civil Rights Act adds a business-services angle. New York and a handful of others have extended coverage in various forms. Most states haven't, and in many the courts interpret state law by reference to Title VII, which means the contractor exclusion imports automatically.

For the underlying protection: the Movement Advancement Project's employment map, with data as of 27 August 2026, counts 24 states plus three territories plus the District of Columbia with laws explicitly prohibiting employment discrimination based on sexual orientation and/or gender identity. Twenty-one states plus D.C. and at least 395 municipalities fully and explicitly prohibit it across employment, housing, and public accommodations. The direction isn't uniformly forward: in 2025 Iowa became the first state in U.S. history to repeal gender identity protections from its nondiscrimination law.

The practical upshot for anyone selling services over the internet: which state's law governs your contract is a real variable you can sometimes control, and almost nobody treats it as one.

Federal enforcement, dated

The enforcement posture shifted in a series of discrete, documented steps, and it's worth knowing them by date rather than by vibe.

January 2025: then-Acting EEOC Chair Andrea Lucas removed gender-identity-related materials from agency websites and documents to align with EO 14168, and the agency stopped using non-binary gender markers in charge intake. Lucas's stated position, which she has repeated including in confirmation testimony, is that Bostock protects against adverse actions — hiring, firing, promotion — based on gender identity, but does not extend to conduct like misgendering or to bathroom and facilities access.

May 2025: a federal court in Texas, in litigation brought by the State of Texas and the Heritage Foundation, vacated the portions of the EEOC's 2024 Enforcement Guidance on Workplace Harassment dealing with sexual orientation and gender identity, holding the Commission had exceeded its authority and misstated Bostock's holding.

22 January 2026: the EEOC voted 2-1, along party lines and without notice or an opportunity for public comment, to rescind the 2024 harassment guidance in full — not just the LGBTQ sections. Lucas stated that rescinding the guidance "does not give employers license to engage in unlawful harassment" and that the Commission remains committed to remedying unlawful workplace harassment.

February 2026: the Commission issued a decision in a matter identified in practitioner coverage as Selina S., which drew a detailed public legal critique from a group of former EEOC officials organized as EEO Leaders, including former General Counsel Karla Gilbride and former Commissioner Chai Feldblum.

Every employment-side law firm writing about this — Littler, Ogletree, Fisher Phillips, Jones Day, Phelps — lands on the same advice for employers, which tells you something: Bostock remains good law, state and local law hasn't changed, and prudent employers should keep doing what they were doing. That's genuinely the consensus among the defense bar.

But for a worker deciding whether to file, the calculus is different from what it was in 2024. Federal charge processing on gender identity theories beyond hire/fire/promote is a materially less certain path. State Fair Employment Practices Agencies matter more than they did. Filing deadlines run 180 to 300 days from the discriminatory act depending on jurisdiction, and that clock does not care about your uncertainty. Lambda Legal's Help Desk and the Transgender Law Center are the two organizations most consistently recommended for triage.

Scams, and a disproportion I can't prove

Job scams are the fastest-growing consumer fraud category in the United States and they are aimed directly at people looking for remote work.

FTC data: reports of job and employment scams roughly tripled from 2020 to 2024, with reported losses rising from about $90 million to $501 million. Task scams — the gamified variety where you're asked to like videos or rate product images, receive small real payments to build trust, then get told to deposit your own money to unlock the rest — went from essentially zero reports in 2020 to about 5,000 in all of 2023 to roughly 20,000 in the first half of 2024 alone. They accounted for 38.8% of job scam reports in that half-year, against 5.6% in 2023, 1.6% in 2022, and 0.6% in 2021. Cryptocurrency is the dominant payment rail: about $41 million lost to job scams via crypto in H1 2024, against roughly $21 million in all of 2023.

It hasn't slowed. Q4 2025 figures put reported job-scam losses at about $150.4 million across roughly 25,002 reports, median loss $2,000. The BBB's 2026 employment scams update found reports roughly doubled year over year, with the typical task-scam loss climbing to about $2,300 in 2025. An April 2026 FTC data spotlight found about one in three people who lost money to a job or business-opportunity scam said it started on social media. The FBI's IC3 put 2025 employment fraud at $362.9 million. Individual losses in the BBB study ran to $57,000, $65,000, $100,000.

Now the honest part. I believe trans workers are disproportionately exposed to this, and I cannot cite a study that measures it, because as far as I can find nobody has run one. What I can point to is the shape of the risk: task scams recruit through unsolicited WhatsApp and SMS messages promising remote, flexible, no-interview work. That pitch lands hardest on people who need income now, who have been screened out of formal channels, who cannot or will not go through a process requiring in-person document verification, and who are carrying the underemployment and unemployment rates cited earlier. Every one of those conditions is documented for trans workers. The intersection isn't.

That gap in the research is itself worth saying out loud, and I'd rather flag it than launder an inference into a statistic. If you write about this, don't invent the number.

The defensive rules are boring and they work. Nobody legitimate recruits you cold over WhatsApp. Nobody legitimate asks you to pay to get paid. Nobody legitimate pays you to like or rate things — that's illegal on its own terms. An upfront background-check fee is a red flag even though it's technically lawful; legitimate employers use consumer reporting agencies operating under the Fair Credit Reporting Act and deduct costs from a first paycheck if at all. Equipment-purchase requirements before a start date are a scam pattern. And if the "sunk cost" feeling has started — if you've already put in money and are being told a little more unlocks it — that is the mechanism working as designed, and the money is gone.

AI ate the on-ramp specifically

The freelance platform economy is contracting, and the contraction hit exactly the rungs that were most accessible.

Upwork ended 2025 with roughly 785,000 active clients, down from about 832,000 a year earlier — a loss of some 47,000, around 6%, the largest contraction in the company's public history. UPWK is down roughly 56% year-to-date in 2026 and about 87% from its July 2021 peak, dropping around 19% in a single session after Q4 2025 earnings in February 2026. Fiverr reported active buyers down 13.6% year over year in Q4 2025. Freelancer Limited, parent of Freelancer.com and Escrow.com, reported FY25 group GMV of A$881.5 million, down 7.1%.

The research on causation is reasonably consistent. Wharton work associated with Manav Raj found measurable displacement of freelance demand in writing, translation, and basic coding following ChatGPT's release. A widely-cited study of an online freelancing platform found monthly writing and editing jobs down about 3% and earnings down about 10% in the five months after launch. A study in Organization Science found freelancers in occupations more exposed to generative AI experienced roughly a 2% decline in contracts and a 5% drop in earnings after the 2022 model releases.

Meanwhile the top end grew. Upwork's Q3 2025 results showed gross services volume up 2% year over year with 52% of that growth coming from AI-related work, and AI-related freelance work crossing $300 million in annualized value by late 2025, with those freelancers earning about 44% more per hour than peers on non-AI projects. Upwork's Future Workforce Index 2026 reported that over a third of skilled U.S. knowledge workers now freelance, up from about a quarter the prior year, and that freelancers using AI earn 34% more per hour — while per-contract earnings for generative AI and creative production work fell 13%, which is the sound of a category commoditizing even as it grows.

Here's why this is a trans-specific story and not just a labor story.

Data entry, basic copywriting, logo design, simple social content — the commodity tier — was the on-ramp that required no interview, no camera, no references, no explanation of a gap in a résumé, and often no more identity verification than a payout account. It was the tier where you could build a work history under your own name without ever having to be seen. That tier is gone or going.

What replaced it is relationship work: retainers, subject-matter expertise, AI workflow integration, ongoing client partnership. Higher paid, more durable, and requiring exactly the sustained personal exposure the commodity tier let you avoid. Video calls. Introductions. A LinkedIn presence with a face on it. Referrals that travel with your history attached.

The ladder didn't just get shorter. The bottom rung, which was the accessible one, is the rung that got removed.

What actually works

Practical, in rough order of how much time it buys you.

Build a name-surface map. Sit down and list every system in your working life, and mark which name each one will display, to whom, and under what conditions. W-9 and 1099s: legal name. Stripe, PayPal, Wise, Payoneer: legal name for KYC, with varying control over the display name a client sees on an invoice. Background checks: legal name plus any AKAs you disclose, and non-disclosure of a prior name can itself be treated as grounds for rejection. HRIS and SSO directories: usually legal name, often propagating into Slack, Zoom, calendar invites, and email display names in ways nobody at the company has audited. GitHub, Calendly, Google Workspace: usually editable. Do this once, deliberately, rather than discovering it one system at a time in front of a client.

Consider an entity. A single-member LLC with its own EIN means the W-9 and the 1099 can carry the business name and the EIN rather than your personal name and SSN. This is not a magic shield — the responsible-party information exists, banking KYC still runs on you, and the specifics depend on your state and how the LLC is structured — but it moves your legal name out of the routine paperwork that passes through a client's AP department. Talk to an accountant. It's a real and underused lever.

Own the client relationship. The strategic version of everything above. Marketplace-mediated work means the marketplace's verification stack, the marketplace's display-name policy, and the marketplace's arbitrary account actions. Direct client relationships mean your contract, your invoice template, your terms. The platforms are contracting anyway; the incentive to build direct is now doubled.

Treat governing law as negotiable. If you're a contractor and the client is flexible on the choice-of-law clause, a state with contractor coverage — California's FEHA, notably — gives you a cause of action that federal law does not. This will feel like an absurd thing to raise over a $6,000 engagement. Raise it on the ones where it's worth raising.

Watch the expiry dates. Know when every identity document you hold expires. A valid passport with an accurate marker retains its validity until that date regardless of current policy, and you are not required to surrender it early. Keep certified copies of any court-ordered name change; they resolve a meaningful share of document-mismatch escalations faster than arguing with a support queue. Understand what your state offers on driver's licenses and birth certificates, because those diverge sharply from the federal position.

Document contemporaneously. The reason I have no case about the SaaS contract is that I have Slack messages and a vibe. Dated notes, saved emails, and a written record made at the time are the difference between a grievance and a claim, and the 180-to-300-day filing windows mean the decision to preserve has to be made long before the decision to file.

Scam hygiene, calibrated. The task-scam pattern is specific and recognizable once you've seen it named. Unsolicited contact, move to WhatsApp or Telegram, small real payments early, then a request for your own money. Never pay to get paid.

If you're hiring

Short list, all cheap to implement.

Label the alias field on your application something other than "former name," and explain why you're asking. Better: check with your screening vendor whether the report can be run without surfacing prior names where it isn't legally required.

Audit what your HRIS pushes downstream. Most companies have never checked whether the legal name in Workday or Rippling propagates into Slack display names, calendar invites, or the email directory. Find out, and give people a way to control it that doesn't require a conversation with a person.

Write a name and gender-marker change workflow before someone needs it, and put it in the handbook. The absence of a process means the process is "come out to HR and hope."

Don't buy demographic inference. If a vendor's product includes automatic gender recognition, emotion analysis, or "candidate fit" scoring off video, the accuracy literature on that is bad and the exposure is worse.

Put nondiscrimination and anti-harassment obligations in your contractor agreements. Title VII doesn't cover 1099 workers; your contract can. It costs you a paragraph.

And if you're a small employer under fifteen people who has concluded that Title VII therefore doesn't apply to you: that's correct, and it is not the flex you think it is.

What I'd say if this were a conversation

The refuge story about remote work is half-true and it was mostly written before 2025. Distance does remove a real category of harm — the hallway, the bathroom, the meeting where someone decides to make a point. Every trans person I know who works online will tell you that part is not exaggerated.

What replaced it is a bureaucratic exposure that's harder to see and much harder to fight, because it doesn't come from a person. It comes from a verification queue, an alias field, an expiring document, a face-match threshold, a choice-of-law clause nobody read. There's no one to confront and nothing to report. Contracts don't get terminated; they get restructured. Accounts don't get banned; they get held for review.

And the legal architecture that was supposed to backstop all this has a hole in it exactly where most internet work sits. Bostock is still good law and it covers employees. Most people doing this work aren't employees.

I don't know how this resolves. The passport litigation is live, the EEOC's composition will change with elections, and twenty-four states plus D.C. offer something the federal government currently doesn't. What I know is that the practical defenses available right now are administrative and boring: know which systems say your name, own your client relationships instead of renting them from a platform, watch your documents, write things down when they happen.

That's a thin set of tools. It's the set that exists.

Figures current as of September 2026. The passport litigation (Orr v. Trump), EEOC enforcement posture, and state nondiscrimination coverage are all actively changing — check MAP's Equality Maps, Lambda Legal, and the ACLU case page before relying on anything here, and consult an employment lawyer in your jurisdiction rather than an article.