Update: the discord server members / friends of Mobius are brigading the comments.

See the model card: https://huggingface.co/Corcelio/mobius

It’s a non-commercial model they want people to pay to use through their API, and won’t allow anyone else to publish the weights, even though they tout the ability to finetune it in the hype post.

Looking deeper into things and it’s using Bittensor to “decentralise AI production”, and it’s using blockchain. Another crypto scam.

It’s quite odd. as a researcher, the claims to cut down on training cost by 2/3rds really stuck out to me, as I would also like to benefit from this advancement. but when you look at how they supposedly achieved this, it’s just another SDXL finetune with 25 million images.

A fun gem from the model card:

• highly suggested to preappenmed watermark to all negatives and keep negatives simple such as “watermark” or “worst, watermark”

A model without any bias shouldn’t really need “watermark” in the negative prompt.

Here’s the license text from the model card:

Mobius is released under a custom license that governs its usage and distribution rights:

Non-commercial use: The model is fully open and available for any non-commercial use. Researchers, students, and enthusiasts are encouraged to explore, modify, and build upon the model freely, as long as they do not use it for commercial purposes.

What is it?

• The EU AI Act is a comprehensive set of rules aimed at ensuring AI systems are developed and used ethically, with respect for human rights and safety.
• The Act targets high-risk AI applications, including those in employment, healthcare, and policing, requiring strict compliance with transparency, data governance, and non-discrimination.

Key Takeaways:

• Prohibited Practices: Certain uses of AI, like manipulative behavior manipulation or unfair surveillance, are outright banned.
• High-Risk Regulation: AI systems with significant implications for people’s rights must undergo rigorous assessments.
• Transparency and Accountability: AI providers must be transparent about how their systems work, particularly when processing personal data.

Why Does This Matter in the US?

• Brussels Effect: Similar to how GDPR set a new global standard for data protection, the EU AI Act could influence international norms and practices around AI, pushing companies worldwide to adopt higher standards.
• Cross-Border Impact: Many US companies operate in the EU and will need to comply with these regulations, which might lead them to apply the same standards globally.
• Potential for US Legislation: The EU’s move could catalyze similar regulatory efforts in the US, promoting a broader discussion on the ethical use of AI technologies.

Banned applications: The new rules ban certain AI applications that threaten citizens’ rights, including biometric categorisation systems based on sensitive characteristics and untargeted scraping of facial images from the internet or CCTV footage to create facial recognition databases. Emotion recognition in the workplace and schools, social scoring, predictive policing (when it is based solely on profiling a person or assessing their characteristics), and AI that manipulates human behaviour or exploits people’s vulnerabilities will also be forbidden.