1/5 🚨 🏃🏽‍➡️Deregulation speedrun today: #AI Omnibus to undergo European Parliament plenary vote, with negotiations between EU institutions starting within hours 🚨

🙅🏽‍♀️ This ‘move fast, break things’ approach shows once again that the AI Omnibus procedure is deeply flawed.

We have been urging EU lawmakers to reject the re-opening of core #HumanRights and safety protects. Yet, the Parliament's position continues to weaken crucial parts of the #AIAct and fundamentally changes the structure of the law.

2/5 Specifically, the Parliament's position:

✅ *️⃣ Reintroduces the vital requirement for high-risk AI systems to register in an EU wide database, including if providers self-assess that their system is not high risk. However, it would still exclude important and readily-available information from this database, which will hinder effective oversight.

3/5 ⛔ Maintains the weakened powers of fundamental-rights bodies. They were bestowed with special oversight powers under the AI Act, but now cannot request necessary information from AI companies, and need to go through an intermediary.

⛔ Undermines the AI Act as a horizontal law without appropriate considerations for consequences, which lies far beyond the purported mandate of ‘technical changes’ of the AI Omnibus.
4/5 ⛔ Keeps the delay of over a year for when the high-risk AI system requirements comes into force, leaving people in the EU without protection from harms by risky AI systems.

⛔ Maintains the Commission's proposal that risks the exposure of people’s sensitive data for ‘de-biasing AI’ with very weak safeguards, despite existing legal avenues such as consent under #GDPR.
5/5 ⛔ Maintains the Commission's proposal of eroding rigorous standards for high-risk AI systems by extending privileges to companies with up to 1700 employees, such as lower fines and lower reporting standards.

✊🏽 We are counting on the Parliament to agree on a final position that centers #FundamentalRights, and for trilogue negotiations to reinforce protections for people.