Seattle Worldcon 2025 did not use ChatGPT to rank more than 1,300 applicants or choose its panels, according to the convention’s later clarification. Human track leads selected candidates first. ChatGPT was then used to help identify potentially disqualifying information about those candidates, while people reviewed the results and made the final decisions.
That distinction corrected the most misleading version of the story—but it did not end the controversy. Critics argued that using a generative-AI system to investigate allegations of harassment, racism, sexism, fraud, and sexual misconduct created unacceptable risks for a convention built around writers, artists, and other creators.
What happened at Seattle Worldcon?
The controversy began with Seattle Worldcon’s disclosure that an artificial-intelligence tool had been used during the vetting of prospective program participants.
The process had three distinct stages:
- Application intake: more than 1,300 people applied to participate on panels.
- Human selection: track leads chose candidates for possible invitations.
- Vetting: a team used a script incorporating ChatGPT to gather online material about selected candidates, looking for information that might make participation inappropriate.
Applicants who were rejected during the initial track-lead selection were not entered into the AI-assisted process. The convention said fewer than five candidates were disqualified during vetting because of previously unknown information, and that no program declines had yet been issued on that basis when the May 6 clarification was published. Worldcon’s clarification said no panelist was selected by AI and no one was excluded solely because of an unreviewed AI output.
Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minutePC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11#1 Best Overall
So the accurate summary is not “AI selected Worldcon’s panelists.” It is: humans selected potential panelists, then ChatGPT was used as an aid in reputational vetting that could affect whether invitations were ultimately issued.
What did the AI-assisted vetting involve?
In its initial April 30 statement, Seattle Worldcon chair Kathy Bond said the process automated online searches that could take 10 to 30 minutes per applicant and saved hundreds of volunteer hours. Organizers said that only a candidate’s name was entered and that humans checked the links and information returned.
The later statement from the chair and program division head disclosed the substance of the prompt. ChatGPT was asked to investigate named individuals for possible “scandals,” including allegations involving homophobia, transphobia, racism, harassment, sexual misconduct, sexism, and fraud. The prompt also requested links or sources.
That wording is significant. This was not simply a conventional search tool returning documents. A generative model was asked to help make a sensitive reputational assessment about real people. Even when a human reviews the results, the model determines which leads appear first and can influence what the reviewer investigates further.
Why did the backlash become so intense?
Reliability and hallucinations
Language models can produce incorrect, conflated, or invented claims. The risk is particularly serious when a system is asked to investigate people who have common names, use pen names, have limited online records, or are discussed in multiple languages.
Worldcon’s original statement acknowledged that the process could return false results. But a warning that errors are possible is not the same as a method for preventing them. A false allegation involving sexual misconduct, racism, fraud, or harassment can harm someone even if a reviewer later determines that it is wrong.
Human review does not remove automation bias
Organizers emphasized that people reviewed the output and made the final decisions. Critics questioned whether that safeguard was sufficient.
There is an important difference between human-in-the-loop and human-controlled decision-making. In a human-in-the-loop process, a person reviews what the system produces. In a human-controlled process, the person independently investigates the matter and treats the AI output as no more than an untrusted, potentially irrelevant lead.
The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →If a reviewer sees an AI-generated accusation first, it may influence the questions they ask, the pages they open, and how they interpret ambiguous evidence. The available public record does not establish exactly how every reviewer worked, whether all candidates received equivalent scrutiny, or whether false positives were systematically logged.
Uneven coverage and identity errors
Name-based research can fail in predictable ways. A system may confuse two people with the same name, miss a pseudonym, overlook non-English reporting, or treat a person’s lack of searchable information as evidence that nothing exists.
The method can also produce unequal results. Some candidates may have extensive public criticism and documentation; others may have little online presence. Well-connected people may have more visible defenses, while allegations recorded in local, private, or offline settings may never appear in search results. These are risks inherent in the approach, not findings that the Worldcon process definitely produced each of these errors.
The cultural contradiction
Many Worldcon members are authors, artists, editors, translators, and other creators. For them, the use of ChatGPT carried symbolic weight beyond the mechanics of the vetting process.
Generative-AI companies have faced sustained criticism from creators and rights holders over the use of copyrighted works in training datasets without permission. The program division head’s apology acknowledged that concern. To critics, a convention celebrating imaginative work had adopted a tool associated with disputes over the treatment of that same work.
Privacy and consent questions
Worldcon said only names were entered and that an outside expert found the privacy protections adequate for the process described in the April statement. That addressed one narrow question, but it did not settle every concern about consent, data retention, provider policies, or governance.
The public record reviewed here does not establish a privacy-law violation. The unresolved issue is whether participants were clearly told that generative AI would be involved and whether the convention had a sufficiently specific policy for handling such searches.
How the public explanation changed
The wording of the explanation evolved during the controversy. The initial statement referred to “using AI tools in our vetting process for program participants.” Subsequent clarification stressed that ChatGPT had been used “only in one instance” and only to help discover material for review after panelist selection had occurred.
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Fix the driver behind crashes, sound loss and screen glitches3Clear out junk files and repair common Windows errorsOn May 2, Bond apologized and said the first explanation had been incomplete and flawed. The problem was not only that readers could misunderstand the technical workflow. The initial statement foregrounded efficiency and volunteer labor before addressing questions about trust, consent, bias, and the consequences of using an unreliable system to surface serious accusations.
Narrowing the role of the tool corrected the claim that AI had chosen the panels. It did not answer the broader question of whether generative AI was an appropriate instrument for high-stakes reputational screening.
Timeline of the Worldcon AI controversy
- April 30, 2025: Seattle Worldcon disclosed that an LLM had been used in vetting program participants. Organizers said only names were entered and humans reviewed the results. (Official statement)
- May 2, 2025: Bond apologized and said the first explanation was incomplete. (Apology and response)
- May 3, 2025: Futurism published “Worldcon Is Getting Eviscerated for Using AI to Select Panelists,” reflecting the broader public framing of the dispute. (Futurism)
- May 6, 2025: Worldcon clarified that humans selected potential panelists, disclosed the prompt, promised to redo the AI-assisted vetting without generative AI, and offered full or partial refunds. (Official clarification)
- May 13, 2025: The chair said the convention was still recruiting people for the re-vetting team and awaiting responses from outside auditors. (May 13 update)
- August 13–17, 2025: Seattle Worldcon took place in Seattle. (Seattle Worldcon)
- February 2026: WSFS business-meeting minutes referred to the disclosure and apology during a later discussion of generative AI. (WSFS minutes)
How many applicants were actually affected?
The figure of more than 1,300 applicants is easy to misread. It does not mean that every applicant was placed into a ChatGPT screening system.
According to the May 6 clarification:
- More than 1,300 people applied.
- Human track leads selected candidates before the vetting stage.
- Applicants rejected at the initial selection stage were not entered into the AI process.
- Fewer than five candidates were disqualified during vetting because of previously unknown information.
- No program declines had yet been issued on that basis at the time of the statement.
The sources do not identify the candidates involved or establish that ChatGPT directly caused a particular person to be excluded. That limit matters: the controversy concerns the design and legitimacy of the process, while the public evidence does not prove a specific AI-generated false accusation or a specific invitation being denied because of one.
Best Value
Were the Hugo Awards decided by AI?
No—Seattle Worldcon explicitly said that generative AI was not used in the Hugo Awards process. The AI disclosure concerned program-participant vetting, not Hugo nominations or finalist selection. (Worldcon’s statement)
The wider crisis also involved resignations from three people in the Seattle Worldcon WSFS/Hugo-related leadership structure. Their timing placed them in the same controversy period, but the available sources do not show that all of those resignations were caused solely by the panelist-vetting incident. Nor should this episode be collapsed into the separate controversy surrounding the 2023 Hugo process.
What did Worldcon promise to change?
Worldcon announced several corrective measures:
- Redo the AI-assisted vetting with new volunteers and without generative AI.
- Invite experienced Worldcon programmers to audit the process.
- Offer full or partial membership refunds.
- Review internal communications, staffing, and organizational structures.
- Consider additional oversight for the chair and leadership team.
- Provide further updates to members.
The May 13 update showed that re-vetting and outside auditing were still being organized. As of the available record through August 18, 2026, the sources establish the promises and early progress reports, but do not conclusively document the final completion of every remediation step, the audit’s conclusions, or the total number of refunds issued.
What a defensible vetting process would require
Worldcon was trying to solve a real operational problem: volunteer-run conventions have limited time, and organizers said conventional research could take 10 to 30 minutes per applicant. Saving hundreds of volunteer hours is a legitimate efficiency goal. It is not, by itself, evidence that the resulting process was more accurate or fair.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
A safer process for high-consequence vetting would begin by defining the relevant standards before searching. “Scandal” is too broad and subjective. A policy should distinguish between an allegation, a documented finding, an admission, a settlement, and a conviction, and should specify which categories actually affect eligibility.
It should also:
- Use ordinary search and database tools only to locate material, not to decide whether conduct is disqualifying.
- Require independent confirmation from primary or highly reliable sources.
- Use at least two reviewers for serious allegations.
- Record the evidence, date, source, reviewer, and reasoning.
- Check identities carefully, including common names and pseudonyms.
- Account for international and non-English sources.
- Provide a correction or appeal mechanism for candidates.
- Document conflicts of interest and inconsistent application of the rules.
- Publish the general policy before applications open.
These safeguards would not eliminate difficult judgments, but they would make the process more transparent, auditable, and contestable.
What remains unresolved?
The public record does not conclusively establish:
- whether the re-vetting was fully completed;
- what the outside audit concluded;
- whether any invitation changed because of the original AI-assisted process;
- whether Seattle Worldcon or later Worldcons adopted a permanent policy on generative AI;
- how many members received refunds.
Those unanswered questions should not be filled with speculation. The documented conclusion is narrower and more useful: Seattle Worldcon used ChatGPT after human selection to help investigate potential problems with prospective participants, faced intense criticism over the risks and symbolism of that choice, apologized, and announced a manual re-vetting and governance response. The incident demonstrated that “a human checked it” is not the same as proving that a process is reliable, fair, or appropriate.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
Recommended Free Tools
Say something about this film