There are a lot of people who are sure they can detect AI generated images. Just look for the weird fingers, they will tell you. Or the reflections in the eyes, or imperfections in regular textures. Other people don't use rules like that, but strongly believe their gut will tell them when an image was AI generated. (I'm one of those people - I deep-down believe I can always tell an AI image based on an indefinable something, or at least I did until someone pointed out to me that an image I'd scrolled past, of the pope wearing a Balenciaga puffy jacket, was AI generated)
There is software that claims to be able to detect AI generated images, but even in their own marketing materials, none claim to have 100% accuracy. And the errors take two forms, both of missing images that were AI generated, and falsely marking human-generated artwork or photos as the work of AI. In the realm of detecting AI-written prose, I have heard that false alarms have been reported by some neurodivergent people, for their writing being too well structured, grammatically correct, or using the em dash.
Balancing those types of errors is called setting the detection threshold, in the field of detection theory: the study of how to make decisions in the fog. It was originally formalized for radar technicians in World War 2, who needed to think about the uncertainty of deciding whether a blip on the radar scope was an enemy plane. It's useful across many fields, and I think it's useful for organizers handling consent issues including reports of consent violations. The awful uncertainty for organizers can take many forms: which of two versions of events is true; is someone a dangerous creep or just awkward; was something a one-time mistake or part of a deliberate pattern; can someone become ok with feedback on their actions or not. If an organizer does it long enough, they will face all those kinds of decisions, and that means choosing their threshold.
"It is better that ten guilty persons escape than that one innocent suffer." is a statement about detection threshold, maybe appropriate for the legal system which is backed by threats of prison. "Always believe survivors", is another one, although few if any organizations ultimately follow this: acting on all consent reports as literally true. Recently I have seen statements such as "[believing victims] does not mean that everything anyone who claims to be a victim of abuse is guaranteed to be true, or the full story." (link) and by someone else, "You can't 'Believe Survivors' before you've decided who the Survivor is".(link)
Detection theory says that thresholds can't be set without thinking about the consequences of both kinds of mistake, misses and false alarms. There is no objectively best place to set the threshold. Letting a bomb through an Xray security device is such a disastrous miss that airports are set up to have thousands of false alarms a day, whereas accusing a scientific article of being falsified makes such waves that it is usually only undertaken with a stack of overwhelming evidence - false alarms are rare. Banning the wrong person, besides being unjust, can tear apart a community, and even act as a tool of oppression against marginalized groups. On the other hand, though the destruction is less immediately obvious, failing to ban someone can mean allowing a predator or creep to operate freely in a community, with long-lasting harm - again, often starting with the most vulnerable groups.
Some have written that it's important that all consent reports get a full-scale investigation, or at least that "both sides of the story" are formally collected. The hope with this is to thin the fog, make decisions with more certainty. I'm neutral about this - even if they have the time and energy to conduct an investigation worthy of the name, organizers don't have training in investigating, or access to resources like CCTV footage, phone records, or forensics. And because of the nature of most consent violations, so often you will deadend at two conflicting statements by the people involved, on something that happened in a private room with no one else there. There are even cases where hypothetical hidden camera footage of the incident wouldn't settle it, because the important part happened on the inside, or in the context of a long relationship. And determining someone's intent or future safety is even more difficult. I think asking a few questions can be a good policy, but even if the investigation is extensive, fog will still remain.
Everyone entering a consent decision-making role needs to know about the fog - not just in kink, because every group will eventually face consent violation reports. They need to think in advance about how they will deal with it.
Differing positions are ok! It is a defensible decision to only ban based on the equivalent of an AI hand with 7 fingers, a high detection threshold. That's what I hear when organizations say things like, "in 10 years we've never had to ban someone, don't be the first!" or "we can only act on incidents that happened at our event". I don't really get that, but the message is clear. Conversely, a group could announce that it bans on a hair trigger, a low detection threshold. The vast majority are going to land in between, with different long term outcomes. The important thing is to communicate that threshold truthfully, so potential attendees can make an informed choice. (one of the worst ways it can land is a group advertising fierce protections, which then chickens out when first faced with uncertainty and does nothing - in particular, "zero tolerance" is always an empty promise)
The fog is one of the main reasons caring organizers burn out. The ambiguity, the worry about being manipulated, sometimes having to wade into a complex multi person situation with abuse accusations in multiple directions. And in general knowing you have likely made mistakes, of both kinds, both of which can cause harm. In most cases they will never know for sure what was the right call, but they have to make it. There is no safe or neutral choice when handling consent violation reports, there is only thresholds.
That's just part of being an organizer. Looking into the fog, and making decisions with your whole heart and mind. Beware of people who say they can always see through the fog. Beware of people who say you don't have to stare into the fog, you can follow some simple rules to guide you through every time. Beware of people who, if the fog around a person or event is eventually cleared away, say you should have been able to see clearly before.
If you want to organize, prepare to live in the fog.
