Key Stable Diffusion Researchers Leave Stability AI as Company Flounders

Reactions to Stability AI’s Troubles

• Some express schadenfreude, seeing the company’s struggles as deserved due to perceived harm to artists.
• Others emphasize that, despite business problems and a controversial CEO, Stability released powerful open models that prevented this tech from being locked behind proprietary paywalls.
• There is curiosity about whether an undisclosed scandal is driving staff and investor departures, but nothing concrete in the thread.

Ethics and Legality of Training Data

• Many argue current AI is built on large-scale “piracy”: scraping copyrighted works without permission or compensation, then selling access.
• Counterpoint: humans also learn from unlicensed art; if models are sufficiently “abstract,” some argue licenses shouldn’t be needed.
• Others reject this analogy, stressing that bulk downloading copyrighted works for commercial products is unlike human learning.
• Debate over remedies: suggestions range from making such models public domain to destroying infringing models entirely.

Fan Art, Fair Use, and Double Standards

• One camp says much of the “art community” already lives on infringing fan art, so its outrage at AI training is inconsistent.
• Others respond that fan art is often noncommercial, low-impact, or tolerated by rightsholders, whereas generative models are commercial products that directly compete with original creators.
• There is disagreement on how “commercial” fan art really is and whether it meaningfully competes with original IP.

Economic Impact on Artists

• Many see AI as uncompensated value extraction from already-precarious artists, worsening inequality and “renting humanity’s mind back to us.”
• Others frame job loss as another wave of automation; focus should be on retraining and safety nets rather than “nerfing” technology.
• Some note AI companies still depend on ongoing human artistic output; overexploiting this commons may backfire.

Open Source vs Closed Models and Stability’s Role

• Stability is praised for releasing usable open models (SD 1.5, SDXL) that power rich community ecosystems and local workflows.
• Concerns: monetization struggles, moves toward more restrictive licenses, and competition from forks that surpass the original.
• Discussion on funding: training remains expensive, which pushes many “open” players toward partial closure or “openwashing.”

Media and Source Skepticism

• Several commenters distrust Forbes and legacy media generally, seeing paywalled pieces and “hit articles” as biased or pay-to-play.
• Use of archive links is debated: helpful for access, but conflicts with HN norms and searchability.