⚡️ EU Finalizes AI Act

PLUS: FTC examines Microsoft's investment in OpenAI

Good Morning. Late-night discussions among EU lawmakers have finally borne fruit with a landmark political deal on a comprehensive, risk-based framework for AI regulation—a first on the global stage. Let’s dive right in.

Today’s Highlights:

  • FTC examines Microsoft's investment in OpenAI

  • OpenChat innovates LLMs with advanced data evaluation

  • Mistral AI's latest model sails the torrential seas

DEEP DIVE

EU Breaks New Ground with World's First AI Regulation Framework

In a monumental stride for technology governance, the European Union has clinched a historic deal on the first-ever comprehensive AI regulation, aptly named the AI Act.

"The EU's AI Act is the first-ever comprehensive legal framework on Artificial Intelligence worldwide...By focusing regulation on identifiable risks, today's agreement will foster responsible innovation in Europe."

— Ursula von der Leyen, President of the European Commission

The legislation, sparked by the increasing prevalence of GAI, marks a critical transition towards ethical and responsible AI development, safeguarded against its potential perils.

The AI Act introduces a structured risk-based framework that categorizes AI systems into three principal tiers: minimal, high, and unacceptable risks. It meticulously outlines that low-risk AI applications will encounter minimal regulations, while high-risk ones used in sensitive areas like medical devices and criminal law enforcement will undergo rigorous scrutiny.

Not just a regulation, but a global benchmark

With this landmark legislation, the EU positions itself at the forefront of a global race towards thorough AI regulation. Surpassing initiations by nations like the US, UK, and China, the EU's regulation is pioneering, setting the stage for global standards in AI technology.

Debate over GAI

Heated discussions ensued over the governance of GAI systems, such as ChatGPT, due to their societal and occupational implications. A consensus emerged, mandating transparency and scrupulous technical documentation for such technologies, with enhanced oversight for models that bear "systemic risks."

The penalties for non-compliance are substantial

Fines ranging from €7.5 million or 1.5% of global turnover to €35 million or 7% of turnover, conditional on the nature of the infringement and the size of the entity, could be imposed for violations. Notably, there's leniency for SMEs and startups regarding the scale of punitive measures for offenses.

The implementation schedule for the new law is staggered with a six-month post-adoption period for prohibited use cases, 12 months for transparency and governance rules, and 24 months for the full suite of requirements. This timeline suggests the full implementation of the EU's AI Act might not occur before 2026.

PUNCHLINES

Duly Noted: Google’s AI note-taking app NotebookLM is now generally available to all US users.

Artificial Influence: As the FTC scrutinizes Microsoft's hefty investment in OpenAI, the lines between collaboration and control get blurrier.

The Musk Touch: X begins rolling out its rebellious chatbot Grok to subscribers.

Pixel-Perfect Peril: AI influencers are reportedly destroying women’s self-image.

The Pirate's AI: Mistral AI's latest LLM hits the net via torrent, seeding a new era for open-source models.

TLDR

OpenChat framework enhances language AI: Researchers unveil OpenChat to advance LLMs by weighing data quality using C-RLFT without explicit labeling. This scalable method enhances model performance even with mixed data and is freely available for commercial use.

University of Sydney innovates multitasking chip: University of Sydney researchers create an integrated circuit chip with notable 15 GHz bandwidth, foresighted to boost microwave photonics. The chip, married with electronics and photonics features, could revolutionize communications, with potential in 6G/7G networks and advanced radar systems.

Inflection AI's Pi now on Android with a personal touch: Pi, the AI chatbot by Inflection AI, co-founded by DeepMind and LinkedIn visionaries, hits Android. Aiming to surpass ChatGPT with voice interactions and a user-friendly persona, Pi offers 24/7 company and personalized advice in 35 countries.

AI gets closer to human-like behavior with 'CHOIS': CHOIS, an AI system from Stanford and Meta, generates lifelike 3D human movements from textual instructions, presenting future possibilities where AI can choreograph complex animations or instruct robots using natural language.

TRENDING TOOLS

Strut AI: Revolutionize writing with an AI-powered collaborative workspace

🚨 Openlayer: Receive alerts when your AI systems stumble, via Slack or email

🏋️ Dumbbell: Transform your workout with motion tracking and automatic rep counting

🎬 Animaker AI: Create AI animation and live-action videos

🤖 GPT Wrapped: ChatGPT year in review recaps your usage in 2023

That’s all for today—if you have any questions or something interesting to share, please reply to this email. We’d love to hear from you!

P.S. If you want to sign up for the Supercharged newsletter or share it with a friend, you can find us here.

Reply

or to participate.