OpenAI's Sora looks Magical

Sora (OpenAI’s prompt to video model) is here…and by the looks of it, Hollywood video editors seem to have some steep competition.

The world took note of Sora this week- OpenAI’s new prompt to video model

Sora can do:

-Text to video

-Image to video

-Video to video

Combine Sora with eleven labs voice overs, and an Apple Vision Pro? We’ve got our own personal Metaverse right there at that point!

​​Simply witness these demos:

Input Prompt: “A stylish woman walks down a Tokyo Street filled with warm glowing neon and animated city signage. She wears a black leather jacket, a long red dress, and black boots, and carries a black purse, she wears sunglasses and red lipstick. She walks confidently and casually. the street is damp and reflective, creating a mirror effect of the colorful lights, many pedestrians walk about”

We are now creating actors and movie scenes in natural language- a marvel. If Hollywood writers had a strike over AI text generation, it’s interesting to consider what they would do in this scenario…

The compute must be insanely expensive- but as with all things in AI, things will become must more cost efficient exponentially.

AI really is like electricity, a technological marvel that is leading to exponential technological leaps.

Now, onto AI cybersecurity discoveries

A paper just came out, stating that AI Agents can autonomously hack websites. 

Essentially, these LLM agents are capable of working independently to perform attacks such as SQL injections- the key is independently. Being independent means this can be done at a much higher rate, faster, and can potentially be a humungous issue.

But, it does not warrant a potential slowing down of ai development- if anything, it should ignite a sense of urgency for AI developers to develop counter security measures.

In fact, a new breed of AI startups are working on this problem as we speak!

Fight fire with fire.

But this is just the beginning…

AI Movies and AI Agent hacks

Yin and Yang.

Anyways, enjoy the weekend.