Currently, many social media sites tried/are trying to implement NFTs, Crypto and AI support without taking in consideration their own user base. On top of it wasting large amounts of electricity, they also incentivize scams and undermine real artists.
Right now we are experiencing an unprecedented rise of bot activity powered by AI, reaching up to 49,6% of total internet interactions. This cheapens the internet experience, turning what is supposed to be a web of people connecting and sharing experiences, into lifeless machines focused simply on increasing numbers and farming reactions.
This moderation list aims to contain all users that engage in such content. This way, you can have the option to mute or block them all at once, greatly improving the quality of your Bluesky feed. Note that using this list is optional, opt-in and it is always possible to stop using it should you change your mind.
This Bluesky Moderation List is updated by collecting posts through a Feed that detects certain keywords, manually analyzing the users and finally adding them if they're found to actually engage in NFT/AI/Crypto content.
This Feed only detects the words, but has no power to add users into the Moderation List in an automated way. It also cannot analyze the text in user bios — it always requires human analysis for a user to be added into the moderation list. We also work by adding and removing users via request.
Our human analysis checks for these elements:
1. Repetition.
AI art posters will usually post several iterations of the same subject in quick succession, in a way that would be difficult for a human artist to reproduce with human effort;
2. Quality.
AI art often has mangled text, hands and inaccurate textures. Lately this has become harder to detect due to AI improvements, so we will only flag them if we are sure;
3. Loudness.
AI/Crypto/NFT supporters will often post about it constantly and passionately. Most of them generate images in order to farm engagement. If a lot of such posts are found on their feed then it will have a greater chance of being added into the list;
4. Links to NFT sites.
If an artist posts their NFT site links then they will be added into the list, regardless if the art was AI-generated or not. Many artists do not desire to see NFT content at all.
1. NFT avatars, honest or ironic.
NFT content are frowned upon many artists. People with "crypto punk", "bored apes" and similar avatars are blocked by the list regardless of their intention.
2. Makes NFTs, even if it is human-made.
Many artists do not like NFTs and it is largely considered to be a scam. Therefore, dealing with NFTs in any way (producing and selling them) will be blocked by this list.
3. Make, facilitate & advertise crypto TXNs (transactions).
Posts that seem to push ads to purchase or invest in cryptocurrency. Advertisements in general are irritating and makes socialization harder in a social media site. This also applies to people selling their crypto art, posting about purchasing crypto material, and sending unprompted DMs to purchase crypto material.
4. eth/btc on name/bio.
Most people with "eth", "btc" and similar suffixes are bitcoin advocates and thus promptly added in the list. Users with "bitcoin symbols" as their avatar or banner will also be blocked. Of course, false-positive words like "Ethan" and "Ethernet" are not added into the list.
5. Make and/or post AI "art", honest or ironic.
People whose main Bluesky interaction is to post AI-generated "art" constantly, regardless of intention. This also detects people who generate AI "art" for fun or "for the memes". AI generation companies consume larger and larger quantities of electricity every day and both "meme AI usage" or "serious AI usage" contribute to this bleak scenario.
Things considered "AI art" are: pictures, videos, music and text that have been generated, in part or wholly, by a computational algorithm such as Midjourney, OpenAI, Gemini, Grok, ChatGPT, Suno etc., with no input from a human other than the initial prompt.
6. Collect AI "art".
In order to feel more important, many AI and NFT art advocates describe themselves as "art collectors". Those are also detected by the list.
7. Mainly follow shill accounts.
People within this circle tend to communicate with their peers. Often, they push each others' "art projects" to seem that a lot of people are interested in it, but in reality they're being boosted by the same group constantly.
8. Are hostile towards real artists.
People who claim that AI is the future often belittle artists who take their time in their craft. For example, calling people who dislike AI-generated "art" a "fascist" or "" is not a good comeback response. This foul behavior is also detected by the list.
9. Want to buy art for NFT projects.
Some people ask artists if they can purchase art for NFT purposes. Artists in general do not want to engage in such activities, and therefore they can be added in the list.
10. Use unethical ways to scrape data for AI training, and its supporters.
People who scrape and publicize data for Machine Learning and AI are to be immediately added in this list for not being ethical in doing so. All proper AI research within a group or community must collect consent from every participant, set up a formal way to ask for data removal and its proof, as per Bluesky Developer Guidelines, and according to the RTBF (Right to be Forgotten) as well as many international laws regarding personal data such as GDPR (Europe), LGPD (Brazil) etc.
Just because Bluesky is a public place does not mean it is public domain; similarly, one should not record people just because they are strolling on a public park.
All people who like and/or follow the perpetrator are to be added in this list as well.
In order of likelihood, from most likely to become a false positive to less likely:
1. AI scientists and researchers.
They should not be included in the list as they are studying the impact of AI and its technologies, but due to being closely related to the Feed's keywords, they will be the most vulnerable to the Feed and our own analysis.
2. AI Scam detectors.
Some content creators work diligently to expose scams by AI and Crypto con men. However they might be caught by the Feed due to similar usage of the keywords.
3. News outlets that happen to post articles about AI, NFT, Crypto.
Legitimate news outlets may talk about AI, NFT, Crypto and their implications. This is an informative piece and therefore shouldn't be added into the list.
4. Blockchain technology.
Blockchain is closely related to cryptocurrency, but on its own, it is just a type of technology. The bulk of scams and predatory monetization happens directly with NFTs and cryptocurrency, not the ledger system.
5. People who post AI, NFT and Cryptocurrency content in other sites, but not Bluesky.
We cannot add people in a Bluesky Moderation list for things that they did on other sites. We must have irrefutable proof that the user posted those items on Bluesky, for safety purposes.
6. Cryptography.
As a math/tech field, this should not be included in the list, but may appear in the Feed due to the text string "crypto".
7. Cryptozoology.
Obviously has nothing to do with Cryptocurrency, but may crop up in the Feed due to the text string "crypto".
8. Workers who use AI as ethical tools.
Many workers use LLM and AI trained ethically for their everyday job. For example, photographers that use AI to remove people from pictures; scientists that use AI to detect cancer cells; safety tools that detect potential bad actors via pattern recognition and automation. Those are not actively trying to shill AI or replace human interaction and workforce. AI tools developed especially to facilitate these kinds of work should not be included in this list.
9. Videogame AI.
By this definition, "Videogame AI" refers not to Machine Learning or Generative AI, but things like "coding behavior of enemies and game logic". This type of AI is fundamentally different from Generative AI.
10. People named "Ai", "Ethan" or similar.
As mentioned on the last item list, names that may trigger the Feed, such as Ethan, Ai, etc. naturally should not be included. Our feed usually knows those are false positives and ignores them, but there's always a non-zero chance that they might end up in the list by mistake.
11. People who don't post AI images as their main type of content.
If the person posted it only once (i.e. it is not their main type of content) then AI Imagery Labeler should be used instead. This is because AI Imagery Labeler is less harsh than a full block that a moderation list provides.
12. People who have some cryptocurrency.
Merely having cryptocurrency does not trigger the Feed, and as long as the user is not being obnoxious about it, they should not be added in this list.
13. People who absolutely have nothing to do with the list.
This is the most unlikely scenario: a complete accident that shouldn't have happened at all. If detected, it must be corrected as soon as possible.
We take full responsibility for any mistakes and once they are detected/reported, we will take immediate measures to correct it. Use the Official Form to request your removal from the List.
IF YOUR CONCERNS INVOLVE THE USERS "ADGIRLMM", "DONKOCLOCK", "PATRIOTBANANA" OR "WINTERFEMINIST", PLEASE NOTE THAT THEY WERE INCLUDED IN THE LIST, CORRECTLY, AS PER RULES OF CHAPTER 3, PARAGRAPH 5 OF THIS TEXT. DO NOT INSIST.