🔭 ChatGPT's agent can now do Deep Research for you

Also: EU just banned AI systems with 'Unacceptable Risk'

Author

Evolving AI
February 03, 2025

Welcome, AI enthusiasts

OpenAI's new Deep Research tool searches and interprets information from the internet using a browser to create a comprehensive report. Starting February 2 in the EU, AI systems deemed 'unacceptably risky' or harmful by the union's regulators are being banned from use. Lastly, AI music startup Riffusion has launched its service in public beta alongside the debut of its latest musical-AI model: Fuzz. Let’s dive in!

In today’s insights:

  • ChatGPT's agent can now do Deep Research for you

  • AI systems with 'Unacceptable Risk' are now banned in the EU

  • Riffusion’s free AI music platform could be the Spotify of the future

Read time: 4 minutes

🗞️ LATEST DEVELOPMENTS

Source: OpenAI

Evolving AI: OpenAI is rolling out a new AI agent for complex research, giving users access to deeper insights and citations.

Key Points:

  • ChatGPT deep research tackles complex inquiries for professional users.

  • It uses a special o3 model to collect and analyze data with clear citations.

  • Results come after 5 to 30 minutes with plans for added visuals and wider access.

Details:

OpenAI introduced a new ChatGPT feature called deep research, aimed at users who need more than quick summaries. It’s designed for detailed inquiries in fields like finance and engineering, as well as for researching major purchases. Deep research pulls from multiple sources, compiles citations, and summarizes its reasoning. For now, it’s available to Pro users with a 100-query limit per month, with broader access expected soon. The feature takes between 5 to 30 minutes to generate results and currently supports only text-based responses. Future updates will add visualizations and expanded data sources.

Why It Matters:

This tool gives professionals a way to get researched, verifiable info for important decisions. Experts in finance, policy, and engineering can use it to back up their work with clear references. Better citations may help improve decision making and workflow reliability. Is this extra clarity enough for critical tasks?

TOGETHER WITH NATURA UMANA
🔉 Natura Umana’s AI earpods evolve with you

Evolving AI: AI today is stuck. You ask, it answers, then forgets — no learning, no evolution.

  • NatureOS changes that. Its AI People remember, adapt, and evolve with you. This is their Humanity Layer.

  • They learn from every interaction, shaping their responses to fit you. And with HumanPods, it's a screen-free experience — just tap and talk.

  • This isn’t just AI. It’s AI that works with you.

Source: Malte Mueller / Getty Images

Evolving AI: The EU's AI Act has reached its first deadline, banning AI systems that pose serious risks.

Key Points:

  • The AI Act sorts AI into four risk levels. The highest risk systems are now banned.

  • These include AI for social scoring, biometric surveillance, emotion tracking at work, and predicting crimes based on appearance.

  • Companies that break the ban face fines up to €35 million or 7% of annual revenue. Full enforcement starts in August.

Details:

The EU's AI Act became law in August. Now, its first compliance deadline has arrived. AI systems that pose an "unacceptable risk" can no longer be used. These include AI for social scoring, biometric tracking, workplace emotion detection, and predicting crimes by appearance. Any company using these AI systems in the EU, no matter where they are based, risks heavy fines. The exact enforcement process is still being shaped, and more rules will come in early 2025.

Why It Matters:

This is the first big move to regulate AI at this scale. Many companies already avoid these AI uses, but the AI Act makes it law. How these rules play out will show if stricter AI oversight spreads to other parts of the world.

Source: Seth Forsgren / Hayk Martiros

Evolving AI: OpenAI believes DeepSeek may have used its models without permission, sparking concerns about data security.

Key Points:

  • Riffusion's AI model, Fuzz, makes music from text, audio, or images.

  • It learns what users like and improves their music over time.

  • Unlike others, Riffusion is free, making AI music easy for everyone.

Details:

San Francisco startup Riffusion has launched a free AI music platform. With its AI model, Fuzz, users can turn text, sounds, or images into songs. The model gets better the more it's used, learning what each person likes. Unlike big names like Google and Meta, Riffusion keeps access open to all. Musicians, including Alex Pall from The Chainsmokers, see real value in it. Early tests show Fuzz can make better music than other AI models using the same input.

Why It Matters:

Riffusion’s free platform lowers the barrier for anyone to experiment with music, from hobbyists to professionals. As AI tools improve, they could change how songs are made, influencing everything from personal projects to commercial releases. The question is — how will artists and the industry adapt?

 👀 Click on the image you think is real

🤖 AI designed computer chips that the human mind can’t understand.

🌱 Google’s X spins out Heritable Agriculture, a startup using AI to improve crop yield.

👨‍💻 Nvidia CEO Jensen Huang says everyone should get an AI tutor right away.

🇬🇧 UK makes use of AI tools to create child abuse material a crime.

✉️ Gmail security warning for 2.5 billion users — AI hack confirmed.

🏢 Microsoft is forming a new unit to study AI’s impacts.

🌍 African schools gear up for the AI revolution.

📈 Trending AI Tools

  • 🔊 CrystalSound - An ultimate real-time noise canceling & voice changer app (link)

  • 📝 Trolly AI - Helps create professional SEO articles 2x faster (link)

  • 🎥 Spiritme - AI-video platform for creating personalized videos with digital avatars (link)

  • 🩺 ScribeMD - Chrome extension for doctors to transcribe and summarize conversations (link)

  • ✍️ Copymatic - AI tool for copywriting tasks (link)

Reply

or to participate.