August 2025 Legal Insights and Updates

On 1 August 2025, the substantive provisions of the Arbitration Act 2025 entered into force across England, Wales and Northern Ireland. The Act amends and updates the Arbitration Act 1996 (“AA 1996”), following a comprehensive two-year review by the Law Commission of England and Wales. The reform process was aimed at modernising English arbitration law and ensuring it continues to meet the needs of parties engaged in domestic and international arbitration.

The Act implements the Law Commission’s recommendations to strengthen the framework of the AA 1996. While the provisions vary in scope, they reflect a drive to clarify procedural rules, refine the powers of tribunals and courts and reinforce the efficiency and reliability of arbitration in England and Wales.

The commencement of the Arbitration Act 2025 represents a significant milestone in the development of English arbitration law. The reforms are expected to enhance England and Wales’ reputation as a leading arbitration hub, providing greater clarity for users and aligning the framework more closely with international best practice. Parties engaged in arbitration should familiarise themselves with the amendments, as they may have practical implications for the conduct of current and future proceedings.

In RTI Ltd v OWH SE iL [2025] EWHC 1945 (Comm), the English Commercial Court summarily dismissed a challenge to an arbitral award under section 68 of the Arbitration Act 1996 (“AA 1996”). The application, which sought to set aside or remit the award on the basis of fraud, was held to be out of time.

The claimants, RTI and Rusal, argued the award was obtained by fraud because O, the defendant, had allegedly failed to disclose documents relating to the service of default notices. They contended the cross-examination of O’s witness indicated the existence of undisclosed communications. Their application was based on section 68(2)(g) AA 1996, but was filed outside the 28-day period set by section 70(3). To overcome this, they argued their earlier request to the tribunal to admit an irregularity constituted an “available arbitral process of appeal or review” under section 70(2), which would defer the time limit for a court challenge.

Butcher J rejected this argument, holding that such a request was not a process of appeal or review. It did not involve a different arbitral body reconsidering the award, but was limited to the tribunal deciding whether to admit or decline to admit the irregularity. The court also refused to grant an extension of time, applying the principles from Terna Bahrain Holding Company WLL v Al Shamsi [2012] EWHC 3283 (Comm). The delay was substantial; there was no adequate explanation for it; and neither O nor the tribunal contributed to it. The application was also weak on its merits.

The judge stressed that section 68 AA 1996 sets a “high threshold” for intervention, limited to “extreme” cases where the tribunal has gone so wrong in its conduct of the arbitration that justice demands correction, citing Czech Republic v Diag Human SE [2024] EWHC 503 (Comm). On the facts, there was no evidence of undisclosed documents, no basis to show any such documents would have been disclosable and no proof of fraudulent non-disclosure.

The judgment highlights the strict approach taken by the English courts to the 28-day time limit for section 68 challenges. Parties cannot rely on applications to the tribunal as a means of extending time and speculative allegations of fraud will not satisfy the high threshold required for relief.

A significant shift towards greater transparency in the judicial system is set to take place in the Commercial Court this October. Mrs Justice Cockerill announced that witness statements, expert exhibits and other essential documents to understand proceedings will be made publicly available. These materials will be accessible through the public-facing section of the CE File electronic filing and case management system.

The move aims to address the “transparency gap” identified by Mrs Justice Cockerill. The decision aligns with the judiciary's ongoing efforts to enhance access to court documents following concerns about confidentiality, timing and costs. This initiative follows the Supreme Court's landmark ruling in Cape Intermediate Holdings v Dring [2019] UKSC 38, which emphasised that open justice requires granting access to documents placed before a judge unless there is a valid reason to withhold them.

From October, documents such as skeleton arguments, witness statements, expert reports and key materials deemed critical by the judge will be made public. However, confidential information will be redacted through a “file modification order” in a relatively informal process, aiming to strike a balance between transparency and confidentiality. This pilot scheme will be applied to Commercial Court hearings in public but will not extend to cases involving litigants in person who are not on CE File.

While the pilot scheme will be closely monitored, it is yet to be determined whether similar practices will be adopted across other jurisdictions of the court, including the Family Division. The initiative is part of a broader effort by the judiciary's Transparency and Open Justice Board, established last year, with plans likely to include live-streaming proceedings in the Administrative Court and providing easier access to case citations in judgments.

This move may raise concerns about the potential misuse of documents, especially regarding the use of court materials to train artificial intelligence systems. However, it is believed that the benefits of increased public access to judicial proceedings outweigh the risks, particularly in a court known for its “sophisticated and engaged user profile”.

The pilot will be implemented subject to ministerial approval, with further feedback anticipated from stakeholders on how best to balance transparency with privacy concerns in the evolving legal landscape.

Belgravia Law is actively monitoring developments related to the CE-File system and the transparency initiatives in the Commercial Court.

On 4 July 2025, the General Scheme of the Arbitration (Amendment) Bill 2025 (the “Bill”) was published, proposing significant changes to the Irish Arbitration Act 2010 (“AA 2010”). If enacted, the new legislation will be titled the Arbitration (Amendment) Act 2025 and will enable Ireland to give effect to investment protection agreements to which the state becomes a party.

Investment protection agreements allow investors to pursue binding arbitration as an alternative to domestic court proceedings in the host country. The purpose of such agreements is to provide a faster and more neutral method of resolving disputes.

The Bill follows the Irish Supreme Court’s decision in Costello v Government of Ireland [2022] IESC 44, which ruled that Ireland could not ratify the EU-Canada Comprehensive Economic and Trade Agreement (“CETA”) without legislative amendments.

Key features of the proposed amendment include:

• Empowering the Minister for Foreign Affairs and Trade to prescribe, by Order, an investment protection agreement to which the AA 2010 will apply.

• Providing arbitral awards made under such agreements may be enforced in Ireland as if they were New York Convention or ICSID Convention awards.

• Clarifying the Irish High Court will refuse enforcement of awards would “materially compromise the constitutional order of the State or the autonomy of the legal order of the European Union”.

If enacted, the legislation will pave the way for Ireland to ratify CETA and other free trade agreements with countries such as Singapore, Vietnam, Chile and Mexico, creating new trading opportunities while offering investors greater security and legal certainty.

An ICSID arbitration tribunal has ordered Argentina to pay more than USD 700 million in damages to US energy company AES, finding that the measures imposed in Argentina’s electricity sector breached the US-Argentina bilateral investment treaty (“BIT”). The award, issued in The AES Corporation v The Argentine Republic (ICSID Case No ARB/02/17, 30 May 2025), is one of the most significant rulings against Argentina in the context of its early-2000s economic crisis.

AES had invested heavily in Argentina’s electricity generation sector in the 1990s. During the 2001–2003 financial crisis, Argentina introduced a range of regulations, including capping spot prices, restricting new power purchase agreements, withholding payments to generators and forcing reinvestment of withheld funds into new capacity. AES argued these measures undermined its investments and violated treaty protections.

The tribunal held that Argentina’s measures were arbitrary and breached the fair and equitable treatment (“FET”) standard, noting the basis for the pricing and regulatory framework was unclear and inconsistently applied. It also found that the measures impaired AES’ ability to manage and enjoy its investments, in breach of Article II(2)(b) of the BIT. However, the tribunal rejected AES’ claim that the FET standard guaranteed a stable regulatory framework and dismissed its full protection and security (“FPS”) claim.

Argentina’s reliance on the BIT’s essential security exception and the customary international law defence of necessity was also rejected. The tribunal found that the measures extended well beyond the period of crisis and continued into Argentina’s recovery, serving to control profitability and compel reinvestment rather than address immediate security needs.

Damages were calculated on the basis of lost gross margins, adjusted for macroeconomic and legal uncertainties. The tribunal awarded AES over USD 715 million, setting the valuation date at the time of the award.

Case: The AES Corporation v The Argentine Republic (ICSID Case No ARB/02/17) (Award) (30 May 2025) (Tribunal: Professor Ricardo Ramírez Hernández (president), Stephen L Drymer (claimant) and Professor Domingo Bello Janeiro (respondent)).

The second edition of Istanbul Arbitration Days will take place between 16–19 September 2025 in Istanbul, Türkiye. This exclusive in-person event is expected to attract more than 2,000 participants, including leading international and local arbitration practitioners, industry experts, academics and students.

Istanbul Arbitration Days is fast becoming a cornerstone event for arbitration in Türkiye and the wider region. The programme will feature world-class speakers addressing current and emerging issues in international arbitration, alongside opportunities for engagement with both local and global business communities. In addition to the main programme, a series of side events will provide valuable forums for building connections and fostering collaboration.

Participants will benefit from extensive networking opportunities with peers from Turkish and international law firms, in-house counsel, arbitrators, company executives and representatives of governmental entities.

Stepan Puchkov, Partner at Belgravia Law and Ceyda Ilgen will attend on behalf of Belgravia Law. We look forward to the insights and connections this event will generate. If you are also attending, we would be delighted to arrange a time for Stepan and Ceyda to meet and exchange views on the latest developments in arbitration.

On 8 August 2025, OpenAI officially launched GPT 5, its flagship model, now available across Free, Plus, Pro, Team, Enterprise and Education tiers. Positioned as the smartest, fastest and most useful AI to date, GPT 5 is built to handle complex tasks like coding, research, writing and healthcare inquiries with greater accuracy and fewer hallucinations than its predecessors.

Key Enhancements

• Dynamic Model Routing: GPT 5 includes an intelligent system that switches between quick-response and deep-thinking modes depending on task complexity. No manual model selection is required.

• Improved Performance: Upgrades include faster response times, higher factual accuracy (up to 80% better in some cases) and extended context windows (up to 256K tokens), bolstering coherence over longer interactions.

• Expanded Capabilities: GPT 5 excels in domains like coding, design, math, science and content creation. Additional improvements span multimodal input handling and the integration of productivity tools.

• Business Adoption: Many organisations, such as Amgen, Morgan Stanley, Lowe’s, are already deploying GPT 5 via APIs or integrating it into workflows, pointing to early enterprise traction.

Despite its advances, GPT-5’s rollout has received a mixed reception. Commentators have described the model’s improvements as incremental rather than transformative, particularly when measured against the hype that preceded its release. While GPT-5 demonstrates clear gains in reasoning, accuracy and enterprise functionality, critics argue these developments fall short of the kind of revolutionary leap expected from a new flagship system. This perception has been especially pronounced in areas such as creative writing, conversational engagement and personality, where some users have noted the GPT-5’s output feels more restrained and less dynamic than its predecessor.

Much of the early feedback has centred on the model’s tone and user experience. Many long-term users have suggested GPT-5 produces technically precise responses but at times colder and more robotic than GPT-4o, which was praised for its more natural, empathetic and engaging style. In response, OpenAI has temporarily reinstated GPT-4o as an option for Plus users, while committing to further refinements to enhance GPT-5’s personality and conversational adaptability.

The launch has also been marred by reports of technical glitches, particularly with GPT-5’s new dynamic routing system, designed to switch automatically between quick and deep reasoning modes depending on the complexity of the task. In practice, some users reported the auto-switcher occasionally misfired, producing either overly brief or needlessly verbose outputs. These shortcomings resulted in inconsistent user experiences during the crucial early days of deployment. OpenAI has acknowledged these issues and pledged rapid fixes, emphasising system stability and reliability remain central to its roadmap.

Taken together, these early criticisms highlight the challenge OpenAI faces in balancing enterprise-level performance with user-focused qualities such as warmth, creativity and natural engagement. For many users, GPT-5’s technical refinements are undeniable. Still, the initial reaction suggests improvements in personality and consistency may be just as critical to the model’s long-term adoption and success.

Strategic Implications

GPT-5 positions ChatGPT as a versatile platform capable of replacing fragmented model suites with one intelligent system. However, OpenAI faces a balancing act: delivering on performance while retaining the engagement and emotional resonance users appreciated. Its enterprise-focused improvements may bolster its long-term value proposition, even if expectations for a leap toward artificial general intelligence were tempered.

On 2 August 2025, new provisions of the European Union’s Artificial Intelligence Act (the “EU AI Act”) entered into force, marking a decisive moment in the regulation of AI. As the world’s first comprehensive legal framework for AI, the EU AI Act sets out obligations for businesses, oversight bodies and Member States. These measures are reshaping how AI is developed, deployed and governed across Europe and they have implications for companies worldwide. This article outlines the key provisions that took effect this month and their significance for legal and compliance professionals.

The EU’s Landmark Artificial Intelligence Act

The EU AI Act represents the most ambitious attempt yet to regulate AI in a holistic manner. Unlike earlier initiatives, which focused narrowly on data protection or sector-specific rules, the EU AI Act takes a broader, risk-based approach to regulating AI technologies across industries. Its scope extends from banning manipulative or exploitative practices, to imposing detailed obligations on high-risk AI systems and creating oversight structures to ensure compliance.

The EU AI Act officially came into force on 1 August 2024, but implementation has been deliberately staggered over a three-year period. This phased approach reflects the complexity of AI technologies and the need for businesses and regulators alike to adapt. The first year saw bans on unacceptable AI practices and the creation of codes of practice for providers and deployers. Now, with the 2 August 2025 milestone, the EU AI Act moves into a more operational phase, introducing enforcement powers, governance measures and penalty regimes.

Notified Bodies and Compliance Oversight

A defining feature of the 2025 milestone is the role of notified bodies. These are independent organisations designated by Member States to assess whether high-risk AI systems comply with the EU AI Act. These bodies will provide third-party oversight, ensuring businesses meet transparency, safety and accountability requirements.

Alongside this, provisions on governance (Chapter VII), confidentiality (Article 78) and penalties (Articles 99–100) now apply. Most notably, financial penalties for non-compliance are enforceable as of this date, creating strong incentives for companies to align with the framework without delay.

Transitional Period for General Purpose AI

General-purpose AI (“GPAI”) models, trained on massive datasets and capable of serving multiple functions, receive special transitional treatment. Any GPAI model already on the market before 2 August 2025 has until 2 August 2027 to achieve full compliance. This two-year grace period recognises the scale and complexity of adapting such models but also signals that regulation of even the most advanced AI will not be optional.

Member State Responsibilities

The EU AI Act also places significant obligations on Member States. By the August deadline, each Member State was required to:

• Report to the European Commission on the resources allocated to their national competent authorities, with updates every two years.

• Designate and notify their competent and market surveillance authorities, making contact details publicly available.

• Establish and communicate rules for penalties and fines, ensuring consistent enforcement across the European Union.

These obligations ensure the EU AI Act is not merely a legal framework on paper but is actively supported by national infrastructure and enforcement.

Codes of Practice and Commission Oversight

Where codes of practice for GPAI providers are not finalised or are deemed inadequate, the European Commission may step in to issue binding rules. This safeguard ensures continuity and consistency of regulation, even if industry-led approaches fall short. In parallel, the Commission must now begin its annual review of prohibited AI systems, with the first review deadline set for 2 August 2025. This mechanism ensures the legislation can evolve to address emerging risks and technologies.

Looking Ahead

The August 2025 developments mark a turning point in the regulatory landscape for AI in Europe. For businesses, this means compliance must move from planning to action. The risk of financial penalties, reputational damage and market restrictions is now tangible. At the same time, companies that embrace compliance may benefit from greater trust and competitive advantage in an environment where accountability is becoming central to innovation.