The Act gives broad exemptions to open source models, but this does not apply to those deemed to pose a systemic risk. The debate continues in the EU over balancing technological advancement with regulation. AI regulation globally is expected to lag behind the pace of AI development. The recent performance of Mistral's new Mixtral 8x7B model, which surpassed OpenAI’s GPT 3.5, underscores the rapid pace of AI innovation.
Key takeaways:
- The EU AI Act, which requires 'high-impact general purpose AI models' to adhere to transparency standards and 'high risk' models to meet additional requirements, has been passed provisionally but will not take effect until two years after final approval.
- Paris-based startup Mistral, founded by Meta and Google researchers, recently announced a $415 million fund-raise from investors, valuing the company at about $2 billion.
- Mistral had been at the forefront of the EU AI Act negotiations, lobbying against a tiered approach to regulating generative AI that ensured foundation models had guardrails.
- AI regulation, whether in the EU, the US, or elsewhere, will be playing catch-up for a long time to come, as the pace of AI development continues to accelerate.