Washington's Lottery forced to pull site after creating AI porn of lotto user

Overall reaction to the incident

• Many see the lottery’s AI promo as highly predatory: combining gambling marketing with personalized fantasy imagery is viewed as manipulative and abusive.
• Several commenters emphasize that the core problem isn’t AI itself, but the state using psychological tricks to push a regressive “tax on the poor.”
• Others downplay the specific harm, arguing a single topless AI image privately shown to the subject is being blown out of proportion, while opponents stress the lack of consent and government involvement.

Morality of lotteries, gambling, and advertising

• Strong thread that state lotteries (and gambling more broadly) are immoral or at least socially harmful, especially when advertised aggressively to vulnerable groups.
• Counterview: lotteries are framed as entertainment and voluntary revenue for public works; some argue banning them would just drive people to illegal markets.
• Broader critique: modern advertising in general is seen as weaponizing evolved dopamine systems, contributing to obesity, addiction, depression, and constant dissatisfaction.
• Some propose sharply restricting or banning most advertising; others argue ads are necessary for market entry and competition.

AI-specific issues: NSFW behavior and alignment

• Multiple comments note that diffusion models often have NSFW content in their training data; prompts like “swimming with sharks” plus beach/vacation themes can nudge toward bikinis or nudity.
• Discussion of safeguards: base-model choice, negative prompts, NSFW detectors, and their limitations. Many suspect the implementers were inexperienced and skipped or misconfigured these layers.
• Contrast drawn between heavily “aligned” commercial models and lightly filtered or porn-tuned community models; some see this incident as exactly why big vendors censor outputs.

Responsibility, liability, and governance

• Debate over who should be held accountable: the engineer, contractors, decision‑makers who chose cheap/unsafe AI, or nobody because “it’s just a bug.”
• Some argue modern “blameless” cultures and AI rhetoric let organizations dodge responsibility (“it was the computer”). Others insist intent and reasonable precautions matter, not zero defects.
• Concern that organizations treat prompts and “parameters” as if they were hard guarantees, misunderstanding non-deterministic behavior.

Technical and operational points

• Speculation that a Stable Diffusion variant or similar was used, possibly porn‑tuned or insufficiently filtered.
• Some argue not storing generated images reduces privacy and breach risk; others note it hinders investigation and accountability.
• Suggestions include using a second AI as a NSFW filter, but commenters note current detectors have significant false positives/negatives.