While the European Union has extended certain deadlines for high-risk AI systems, the core framework remains firmly in place and several significant provisions become applicable from 2 August 2026. Organisations that interpret the recent changes as a reason to pause their AI governance initiatives risk finding themselves unprepared for the regulatory expectations that are already taking effect.
For boards, senior leaders and governance professionals, the question is no longer whether AI requires oversight. The question is whether sufficient governance arrangements are in place to demonstrate accountability, transparency and effective risk management.
The EU Council recently approved amendments to the AI Act through the Digital Omnibus package, introducing several important timing changes whilst retaining the Act’s risk-based regulatory architecture.
The most significant changes are:
However, the overall direction of travel has not changed. The AI Act remains a comprehensive governance framework designed to ensure AI systems are safe, transparent and accountable.
| Date | Requirement |
| 2 February 2025 | Prohibited AI practices ban and AI literacy obligations apply |
| 2 August 2025 | General Purpose AI (GPAI) provider obligations apply |
| 2 August 2026 | Transparency obligations become enforceable and regulatory oversight expands, including enforcement powers relating to GPAI providers |
| 2 December 2026 | Additional synthetic content labelling and watermarking requirements apply, together with new prohibited practices relating to non-consensual intimate content |
| 2 December 2027 | Compliance obligations for stand-alone high-risk AI systems apply |
| 2 August 2028 | Compliance obligations for high-risk AI systems embedded in regulated products apply |
Many articles discussing the recent amendments focus almost exclusively on the delayed high-risk deadlines. That risks overlooking why 2 August 2026 remains a major milestone.
The Act enters a significant new phase on this date, bringing both additional obligations and increased regulatory oversight.
From 2 August 2026, key transparency obligations under Article 50 apply.
These include requirements to inform individuals when they are interacting with AI systems in circumstances where this may not be obvious and to disclose certain AI-generated or manipulated content. Organisations using chatbots, AI-enabled customer service tools, virtual assistants and synthetic media should already be considering how these requirements will affect their operations.
For many organisations, transparency is no longer simply good practice. It becomes a regulatory expectation.
2 August 2026 is also significant because enforcement activity begins to expand in relation to General Purpose AI (GPAI) providers.
The obligations applicable to providers of GPAI models, including requirements relating to transparency, technical documentation and copyright compliance have applied since August 2025. However, August 2026 represents the point at which regulators gain greater ability to supervise and enforce those obligations.
Even where organisations are not developing foundation models themselves, they increasingly rely upon GPAI technologies through products such as Microsoft Copilot, ChatGPT and other AI-powered platforms. As a result, boards should be considering how they obtain assuranceregarding the governance and compliance arrangements of key AI suppliers.
The requirement to ensure an appropriate level of AI literacy has applied since February 2025 and remains one of the most overlooked aspects of the legislation.
Organisations deploying or using AI systems must ensure relevant personnel understand the capabilities, limitations and risks associated with AI tools. This extends beyond technical teams and may include management, operational staff and board members.
The additional time granted for certain high-risk systems should not be viewed as an opportunity to delay.
Instead, organisations should use this period to establish the governance foundations that will support long-term compliance and responsible AI adoption.
Many organisations cannot readily identify all AI tools being used across the business. Understanding what is in use, where it is being used and who owns it is the starting point for effective governance.
Organisations should evaluate:
Boards should consider whether they have:
These are areas where regulators increasingly expect evidence of governance and challenge.
AI literacy is becoming a governance issue rather than simply a learning and development initiative. Organisations should ensure leaders and employees understand both the opportunities and risks associated with AI technologies.
As regulatory scrutiny increases, organisations should be able to demonstrate:
The ability to provide assurance may prove just as important as the controls themselves.
The recent amendments to the EU AI Act should not be viewed as a delay to regulation. They are better understood as a recalibration of implementation timelines.
What has not changed is the growing expectation that organisations understand where AI is being used, how risks are being managed and who is accountable for oversight.
The organisations that will be best placed to take advantage of AI are unlikely to be those with the most advanced technology. They will be those with the strongest governance, clearest accountability and greatest confidence that they can demonstrate responsible, transparent and compliant use of AI.
The revised deadlines may provide additional time. They do not reduce the need to act now.
One of the biggest misconceptions surrounding AI regulation is that compliance is primarily a technology challenge. In reality many of the requirements sit squarely within governance, risk, assurance and accountability frameworks.
At Bridgehouse, we help organisations establish practical governance structures that enable innovation whilst maintaining appropriate oversight and regulatory compliance.
Our support includes:
Whether an organisation is at the start of its AI journey or seeking to strengthen existing governance arrangements, the focus should be on building sustainable oversight rather than simply preparing for regulatory deadlines. If you would like to explore what these developments could mean for your organisation, please contact us to speak with our team.
We would be pleased to answer any queries or have an informal chat to discuss your possible governance needs.