Insights and updates

From emerging regulation to deal mechanics, we write about the questions founders and investors actually ask — practical analysis you can put to work.

AI Literacy (Article 4, EU AI Act)

AI literacy is the Article 4 duty — applicable since 2 February 2025 — requiring providers and deployers to ensure staff operating AI systems have sufficient skills, knowledge and understanding of the risks, tailored to context and role.

High-Risk AI System (EU AI Act)

High-risk is the AI Act’s heaviest tier: Annex I product AI and Annex III use cases like hiring and credit scoring. What changed with the Digital Omnibus — Annex III now lands 2 December 2027 — and what remains binding today.

Fundamental Rights Impact Assessment (FRIA)

A Fundamental Rights Impact Assessment (FRIA) is the new assessment introduced by the EU AI Act (Article 27) requiring deployers of certain high-risk AI systems to assess the potential impact on fundamental rights before putting the system into use.

General-Purpose AI (GPAI) — Concept

General-purpose AI describes models and systems capable of performing many distinct tasks rather than one narrow function — the category the EU AI Act regulates separately through Article 53 obligations on model providers.

Mixture of Experts (MoE)

Mixture of Experts (MoE) is a neural-network architecture in which the model is composed of many specialised sub-networks (“experts”) plus a routing network that decides which experts handle each input token.

Multimodal AI

Multimodal AI describes AI systems that process and generate across multiple input/output modalities — text, image, audio, video, structured data — within a single model architecture.

Reinforcement Learning from Human Feedback (RLHF)

Reinforcement Learning from Human Feedback (RLHF) is the training technique that aligns LLMs with human preferences — making them helpful, harmless and following instructions, beyond what raw next-token prediction would produce.

Prompt Injection

Prompt injection is a class of attack on systems built around LLMs in which an attacker crafts input (a prompt) that overrides, modifies or bypasses the system’s intended instructions.

KISS (Keep It Simple Security)

KISS (Keep It Simple Security) is 500 Startups’ open-source seed-financing template, published in 2014 as an alternative to Y Combinator’s SAFE. Two variants exist:

MiFID II (Markets in Financial Instruments Directive II)

MiFID II (Markets in Financial Instruments Directive II) is the EU regulatory framework governing securities markets, investment services, market infrastructure, and investment-firm conduct — effective January 2018 as the successor to the original MiFID (2007).