Hi Synapticians!

Who owns what an AI makes?

That’s the big question. Copyright means you own what you create. But with AI like ChatGPT, things get blurry. Is it OpenAI who owns the content? Or the people whose data trained the model?

This debate has been going on since ChatGPT launched, and now it’s back — because of the new image feature. Everyone’s turning their photos into Ghibli-style art. It’s fun, sure, but is it legal?

Personally, I don’t think OpenAI should be allowed to use everyone’s data like that. But let’s be honest — there’s no way to control it now. The train has already left the station.

BTW :

Top AI news

1. AI-generated Ghibli-style images spark copyright controversy
OpenAI’s latest image generator, capable of producing Studio Ghibli-style visuals, has gone viral, sparking a heated copyright debate. CEO Sam Altman even adopted a generated image as his profile picture. However, OpenAI is already facing multiple lawsuits over unauthorized use of copyrighted material. The company is lobbying for AI-generated content to be considered fair use, a stance that divides artists and tech leaders. Meanwhile, OpenAI is nearing a $40 billion funding round and expects significant revenue growth. The future of copyright in the AI era remains uncertain.

2. Amazon’s AI now tailors shopping to your hobbies
Amazon has launched 'Interests,' an AI-powered feature that recommends products based on users’ hobbies and preferences. Users can input natural language queries like 'wooden puzzles' or 'romantic decor under $100,' and the AI filters Amazon’s catalog accordingly. The feature allows for multiple customizable profiles and is currently available to a limited number of US users. It’s part of Amazon’s broader AI strategy, which includes over 1,000 active projects.

3. GPT-4o Gets Smarter
OpenAI has released an updated version of GPT-4o with improved instruction-following, enhanced programming and problem-solving skills, and more creative, intuitive responses. The model now uses fewer emojis and is described as having more 'freedom' in its outputs. It is available via API and ChatGPT for subscribers, with free access coming soon. Early analysis suggests GPT-4o now outperforms Claude Sonnet 3.7, reinforcing OpenAI’s lead in the LLM space.

Bonus. Inside Claude 3.5: Planning, Reasoning, and Multilingual Thought
Anthropic’s new research tool reveals how Claude 3.5 Haiku processes information. The model uses a universal conceptual layer to reason across languages, plans rhymes in advance when writing poetry, and performs multi-step logical inference. However, its explanations often differ from its actual reasoning, mimicking human-like responses. A parallel study by Google shows similarities between LLMs and human brain activity, but also highlights key architectural differences. These findings deepen our understanding of how AI models think—and how they differ from us.

Meme of the Day

Theme of the Week

AI for mental health - Real World Applications

Stay Connected

Feel free to contact us with any feedback or suggestions—we’d love to hear from you !