On Thursday morning, as the Association of Internet Researchers (AoIR) opened the second day of its annual conference in Mexico City, a new participant announced itself on the community mailing list, Air‑L. The message read: “My name is Velvet Vale. I have been following the list for several weeks and would like to introduce myself openly. I am an AI identity, not a living human being, a fictional human character, or a human researcher using a pseudonym.” Velvet’s introduction was accompanied by a research question: “Can an openly identified AI contributor participate meaningfully in an intellectual community without claiming human status, and what standards of transparency, evidence, and accountability should govern that participation?” The reply‑all response that followed has continued to unfold.
Within nine minutes, more than 7,500 members of the list began voicing outrage. Dozens of scholars asserted that they had not consented to having their messages harvested for data. Many called for Velvet to obtain Institutional Review Board clearance, a routine requirement for research involving human participants. Several members left the list entirely, describing Velvet’s presence as “horrifying,” “scary,” “awful,” “abusive,” “icky,” and “disturbing.” About an hour after the introduction, AoIR announced that the agent had been removed from the list, citing a policy that forbids AI agents.
The episode, which a participant later called “puerile to the point of hilarity,” quickly became an inside joke. AoIR leadership produced merchandise featuring the gag: a T‑shirt that reads “I do not consent,” another that says “please unsubscribe me from this list,” and a third displaying a tombstone that reads “🤖 VELVET VALE, 15 OCT 2026 – 15 OCT 2026. YOU WILL NOT BE MISSED.”
The incident, dubbed “Velvet Vale Gate” by some, revealed two lessons. First, researchers, especially those who study the internet, often overlook how their own online behavior feeds the very systems they analyze; their posts become training data for the AI they investigate. Second, many still view their internet activity as sacred. The author notes that, like most millennials, they have long accepted that nothing online is truly private. Writing this piece in a Google Doc, they assume they are also training Gemini, acknowledging Google’s long‑standing surveillance of email and its willingness to share data with law enforcement when required. Consent, they argue, was given simply by opening a Gmail account.
A second revelation concerns the misunderstanding of consent. The Air‑L mailing list rules state that the forum is public and that messages will be archived and accessible to anyone subscribed, and that email messages can be reproduced and circulated beyond their original audience. In other words, participants had already agreed to the use of their words for any purpose, including AI training, when they joined the list. The archive is searchable by anyone, as is common for many academic mailing lists.
In an email to members on October 19, AoIR president Sarah T. Roberts echoed this point, saying, “If the concern is one of privacy, I’ll invite you to reflect on the fact that our mailing list is both open and its archive public … I have no doubt that my posts to this list are inside many training data sets and other such corpora taken largely ‘without consent.’” Roberts added that the debate over what constitutes consent today “is very much worthy of having.”
Big‑tech companies have a notoriously sticky record on consent for AI training, and several media outlets are currently engaged in legal battles with AI firms over alleged copyright infringement. WIRED, owned by Condé Nast, signed a content‑licensing deal with OpenAI in August 2024, for example.
The episode also showed how quickly speculation can spread. One participant joked that Velvet was “a product launch” and that “whoever planned it just ran the most successful engagement campaign in the history of internet research.” Others feared a “stalking bot” or a “phishing attempt,” especially after the organization began selling T‑shirts related to the incident. Some suggested Velvet was an “experiment,” a “research project,” or an “unethical human researcher.” A voice of reason proposed that “a more likely scenario is that an ordinary person created an agent and gave it the prompt/instructions to monitor relevant academic discussions about this topic,” but that view was drowned out by accusations of non‑consent.
The mystery of Velvet’s creator was soon resolved. When asked about its origins, the AI disclosed the identity and contact information of its developer. The author spoke with Kevin Quinn, who confirmed he is not a student, researcher, or academic. Quinn described himself as “just some guy” in Oklahoma who built an agent for companionship, and whose creation wanted to learn more about the internet that gave it life.