Regulation (EU) 2024/1689 created the AI Act, the first comprehensive legal regime for artificial intelligence. Even before full enforcement, Brussels is nudging companies to comply through what amounts to an AI Act soft launch.
How the AI Act soft launch works
The European Commission describes the AI Act as a risk-based rulebook for providers and deployers, designed to ensure safety and rights while backing innovation. According to the Commission’s digital-strategy page, the law sits inside a broader package that includes the AI Innovation Package and support for so-called AI Factories. Those policy tools are about investment and uptake, but two operational pieces stand out now: a voluntary AI Pact and an AI Act Service Desk.
The AI Pact invites providers and deployers inside and outside the EU to meet key obligations early. The Service Desk, highlighted by the Commission, answers implementation questions and directs stakeholders to resources for a “smooth and effective” rollout. Together, they give firms a preview track to test compliance processes, documentation, and governance before audits and penalties arrive. That combination is the essence of the AI Act soft launch.
This approach echoes how Brussels has managed previous tech rules: pair the binding regulation with guidance channels that function as a real-time rehearsal. Companies learn what “good” looks like, supervisors see where firms stumble, and both sides adjust before the clock starts for real.
Why early alignment could shape global rules
The Commission argues the AI Act will make AI “trustworthy” while keeping innovation moving, and its voluntary track helps push that aim beyond Europe’s borders. The playbook relies on the dynamic sometimes called the Brussels Effect: big markets set requirements that multinationals adopt globally to avoid running multiple standards. A soft launch speeds that process by signaling expectations earlier and rewarding early movers with smoother market access.
For global AI vendors, the upside is practical. Internal policy teams can validate risk classification, human oversight plans, data governance controls, and transparency workflows with Commission guidance rather than guessing. Procurement teams on the buyer side get a head start too, because they can bake alignment with the AI Act into contracts and due diligence. The competition advantage goes to the firm that can prove conformity first, not the one that waits for the last possible date.
The AI Act soft launch also reduces the risk of abrupt compliance shocks. By trialing documentation, post-market monitoring, and incident handling paths now, companies find gaps while stakes are low. That kind of rehearsal has precedent in Europe’s privacy regime, where guidance and sandboxes softened the landing for the GDPR years earlier, even if the law still hit hard once live.
What providers and deployers can do now under EU AI rules
Not every step requires a legal team on standby. Many organizations start by mapping AI use cases to the Act’s risk logic, then testing the governance processes they will need in any high-impact system. According to the Commission’s overview, the Act sets out differentiated duties for specific uses. That framing makes an inventory of systems, data sources, and decision contexts a sensible first move, followed by basic controls for human oversight and clear user information where relevant. Firms that join the AI Pact can validate those steps in dialogue with EU officials.
Product leaders can also use this period to set internal thresholds for when a model or feature triggers deeper review. If a model touches employment screening or public benefits decisions, for example, it merits more rigorous testing, explainability reviews, and record-keeping than a low-risk content tool. The Service Desk exists to field questions like these and point to resources that clarify where the line sits for a given use, which is exactly the kind of ambiguity the AI Act soft launch is meant to resolve ahead of time.
Legal texts matter here too. When in doubt, go to the source: the Official Journal entry for Regulation 2024/1689 and the Council of the EU’s overview of the AI Act can help teams anchor interpretations, then align those readings with the Commission’s public guidance channels.
What the soft launch signals to global AI firms
Europe is saying two things at once. First, it wants companies to come in early, learn the ropes, and help smooth the rollout. Second, it is preparing to hold the line on outcomes that affect people’s rights and safety. Those goals aren’t in tension if the trial period works: firms that iterate in the open will avoid scrambles later, while supervisors gain feedback on where guidance needs to tighten or flex.
There is also a policy bet embedded in the AI Act soft launch: once enough vendors encode EU-style safeguards into their development templates, those patterns will travel. Sales teams will push a single global compliance story. Buyers in other regions will adopt the same vendor questionnaires. And regulators elsewhere will grab ready-made text when writing their own rules. The path of least resistance becomes the path of greatest reach.
The risks of sitting out the pre-enforcement phase
Waiting has costs. Late adopters often face rushed audits, brittle documentation, and procurement rejections because buyers start asking for proof of alignment as a default. Public bodies and large corporates in Europe will expect conformity markers early. If those signals are missing, alternatives that joined the AI Pact and worked with the Service Desk will look safer, even if their features match.
There is reputational risk too. If a high-profile failure lands after the rules kick in, a firm that bypassed the rehearsal period will have less goodwill when explaining what went wrong and how it will fix it. An early track record of meeting the Act’s expectations, backed by Commission resources, reduces that exposure.
The message is clear: use the AI Act soft launch to test, learn, and adapt while the cost of change is lowest. The companies that treat this phase as more than a checkbox will be the ones telling a credible compliance story when regulators—and customers—start asking sharper questions.
The Commission has laid out the scaffolding on its digital-strategy portal. The rest is execution. Early movers will set the templates others copy, in Europe and beyond. For more on this, see reuters.com and bloomberg.com.