Demystifying AI Governance

​EU AI Act Demystified: Risk Tiers, August 2 Rules, and Future Requirements

• ​Format: 60-minute presentation and panel, followed by 15-20 minutes of interactive Q&A and networking
• ​Target Audience: Early-stage to mid-stage AI founders, technical leads, product managers, and compliance advisors.

## ​Event Agenda

​Part 1: The Four EU AI Risk Tiers (15 Minutes)
​Every builder must determine their system's regulatory category. Non-compliance carries financial penalties up to 35 million euros or 7% of global annual turnover.
​Unacceptable Risk - Social scoring, subliminal behavioral manipulation, emotion recognition in workplaces/schools. Banned completely in the EU. Systems must not be developed or deployed.
​High Risk - AI in recruitment, credit scoring, medical devices, educational admissions, critical infrastructure.Heavily regulated. Requires risk management, high-quality data auditing, logging, and human oversight.
​Limited Risk - Chatbots, AI coding assistants, synthetic media generation, deepfakes.Active enforcement from August 2. Mandatory transparency disclosures and content watermarking.
​Minimal Risk - Spam filters, AI video game NPC behaviour, basic search and recommendation engines. No mandatory regulatory burden under the AI Act; voluntary codes of conduct apply.

​Part 2: Active Enforcement vs. Future Milestones (20 Minutes)
​Currently Enforced Requirements

• ​User Interaction Disclosures: Chatbots and interactive agents must explicitly notify users that they are interacting with an AI system.
• ​AI literacy: EU AI Act Article 4 statutory literacy training.

​Future Developments and Deadlines

• ​AI Regulatory Sandboxes: National authorities must establish operational AI Sandboxes, giving startups a controlled framework to test high-risk models under legal guidance before commercial launch.
• ​Full High-Risk System Compliance: High-risk standalone AI applications and safety components transition into mandatory compliance oversight over upcoming regulatory windows.

​Part 3: Practical Takeaway: The AI Founder's Compliance Toolkit (20 Minutes)
​Attendees will receive an actionable technical and operational framework:

1. ​Risk Classification Decision Tree: A simple logic flow to categorise your product's regulatory tier.

​
Part 4: Interactive Q&A and Community Discussion (15 - 20 Minutes)

• ​Open discussion covering open-source model responsibilities, edge cases, overlapping GDPR rules, and technical implementation challenges.

Registration link: https://luma.com/1gpeahgp