EU AI Act Compliance: Observability is Key, According to Arun P.

In a recent LinkedIn post, Arun P. highlights the impending deadline for the EU AI Act and stresses the critical need for organizations to prepare for its high-risk obligations. The Act’s enforceable date is August 2, 2026, with potential fines reaching up to €35 million or 7% of global revenue. Arun P. frames this not as a distant future problem, but as a current, “this-quarter problem.”

The Evidence Gap in AI Decision-Making

Arun P. points out a significant gap he frequently observes: most companies currently running AI in production cannot provide regulators with the evidence required to explain how their AI systems make decisions and how those decisions are monitored. “The honest answer at a lot of organizations is that the evidence doesn’t exist in any structured form,” Arun P. states, noting that it is often scattered across various logs, documents, and even relies on human memory.

“Show us the evidence of how your AI made this decision, and how you monitor it.”

This lack of structured evidence poses a substantial compliance risk, according to Arun P. He emphasizes that audit-readiness is not a last-minute task but a continuous process. “Audit-readiness is not a document you write the week before an audit. It’s a byproduct of observing your AI continuously, from today,” Arun P. argues.

Observability as the Path to Compliance

The core of Arun P.’s message is that robust AI observability is intrinsically linked to compliance. He suggests that by implementing effective AI observability from the outset, organizations can naturally generate the evidence needed for regulatory scrutiny. “Get the observability right, and the compliance evidence writes itself,” Arun P. asserts. Conversely, neglecting observability means organizations will either be scrambling to manually piece together evidence under pressure or face significant costs to bridge the compliance gap.

To address this challenge, Arun P. mentions that his company, Block Convey, has integrated compliance evidence directly into their observability layer. This proactive approach aims to alleviate the burden on companies as the EU AI Act deadline approaches.

Preparing for the August 2026 Deadline

Arun P. concludes by prompting businesses to reflect on their preparedness. He asks whether their teams are treating the August 2, 2026, deadline as a future event or if they are already actively working to meet its requirements. This question underscores the urgency Arun P. believes is necessary for effective AI governance and compliance in the face of the EU AI Act.

“Skip it, and you’re assembling it by hand under deadline pressure, or paying for the gap.”

The insights shared by Arun P. provide a clear call to action for businesses utilizing AI, emphasizing that a strategic and continuous approach to observability is not just best practice, but a fundamental requirement for navigating the evolving regulatory landscape.