- within Technology, Wealth Management and Employment and HR topic(s)
- with Senior Company Executives, HR and Finance and Tax Executives
What's next in AI?
- Agentic AI has arrived and marks the next stage in the technology's development. As models begin to equalise, agents are how the technology will embed itself in the global economy.
- Regulatory divergence will continue and fears will grow that businesses will adopt a defensive posture by complying with the strictest rules – potentially missing out on the benefits of lighter regimes.
- Geopolitical tension will begin to determine the AI supply chain and states are wrestling with how to control the movements of hardware and software.
- AI superpowers will determine the capabilities of the technology while less AI-capable nations will focus on safety and fairness in their own jurisdictions.
Predictions around technology are often prone to hyperbole – and AI has been no exception to the hype cycle. But we are entering a new phase in artificial intelligence. Agentic AI has scaled the technology underpinning the first generation of chatbots, while an emerging extraterritorial focus from regulators is creating contestation. A new wave of trade policy is expected to target software as well as hardware, and sovereignty concerns are being dictated by AI superpowers, leaving other nations to compromise on home-grown solutions.
As part of our Code≠Law series exploring the challenges in regulating emerging technologies, our experts share their insights and predictions on the standout trends shaping the next phase of AI development and deployment.
Agentic is how AI will be integrated into the economy – and it's already happening
The next major development in AI will not be a vertical improvement on its capability, but a horizontal integration and implementation phase made possible by AI agents.
"We're witnessing a movement from web-based chatbots to desktop-based apps that can navigate, touch and action all your desktop files," says Sydney-based technology lawyer Raymond Sun. "They will be able to take control of your computer as you wish, based on your objectives - that will be the interface we see and get used to."
Though AI agents are underpinned by existing LLM technology, their ability to interact with one another and execute tasks means we are seeing the early stages of a deeper AI implementation across the economy.
"We're already seeing some pioneering entrepreneurs moving towards autonomous organisations - setting up businesses that run themselves through a series of agents," says our Digital Legal Delivery Director and technology lawyer Alex Amato-Cravero. "We are a long way off this being the case for any sophisticated organisation – and there are many legal hurdles to overcome in terms of corporates truly operating without any human involvement – but we can expect to see a rapid rise in the number of businesses whose underlying operations involve an ‘invisible workforce’ of agents operating with limited human oversight."
The challenge for businesses will be understanding how agentic AI fits into existing governance and risk regimes, but the challenge for all of us is understanding how tasks that have historically been beyond the reach of automation will be impacted by the agentic era.
Businesses will take a high-water mark approach to AI rules – and regulators will expect them to maintain oversight of third-party technology
Governments around the world are drafting regulatory regimes that reflect their political preferences, economic priorities, and domestic AI capabilities.
The result, as explained in our AI regulatory article, is a patchwork of rules that vary from comprehensive legislation to more light-touch and experimental approaches. However, as more rules emerge with extraterritoriality, there could also be a regulatory competition underway between blocs.
For businesses, the emerging fear is that they will struggle to untangle these regional differences and simply comply with the strictest rules to protect themselves – missing out on the advantages of more laissez-faire regimes.
Regulators already expect that organisations maintain oversight of established outsourced technologies – and this is being extended to third-party AI.
"Deployers will run into the same issues that financial institutions encountered when they first started to use cloud services," says Frankfurt-based counsel Thies Deike, whose work focuses on IT matters. "It may take a bit longer with AI, but we will see the same development - over time, certain controls will have to be implemented by the deployer of the technology."
The next wave of trade policy will explicitly target AI and software
AI has become a strategic priority for governments around the world and geopolitics will determine the future of the supply chain. The most high-profile example in recent months has been the trade war between the US and China, with the US restricting access to high-performance chips required for AI training, prompting a response from China on rare earth minerals.
But the standoff is extending beyond hardware. The software and algorithms supporting the technology is agnostic: a model being used in a benign way in one jurisdiction could be used in another to orchestrate a cyber-attack, train drones or create biological weapons.
"That's what the second wave of trade policy around AI will focus on," predicts Singapore-based TMT partner Peter Jones. "But it's incredibly challenging to regulate the transfer of the soft stuff – the information and the data. That's the nature of what we will have to deal with going forward - it's got nothing to do with the hardware."
Sovereignty concerns will require non-AI leaders to compromise
There is a growing divergence between states that see AI as a national security concern and those more focused on ensuring AI tools meet domestic standards around fairness and safety.
"For some, it’s largely driven by the need for national security," says Sydney-based TMT sector head Kwok Tang. "As it becomes integral to finance, energy and security, there is a view that the technology must not be disrupted or exploited through international supply chain issues. So, as much as possible there is a sense that it should be built within the home nation."
You can't just have the latest technology, because it's in the US or China and they're not necessarily exporting it.
Kwok Tang
Partner, Sydney
However, for some nations these concerns must be balanced with what is realistic given AI development is currently dominated by the US and China. For nations with concerns that centre more on ensuring AI is safe and fair in their own market, there will need to be a compromise.
"You can't just have the latest technology, because it's in the US or China and they're not necessarily exporting it," adds Tang. "If you have to build the infrastructure from the ground up may be inferior to a ‘best of breed’ approach. So, there has to be a compromise around a completely sovereign solution and a solution that is deployed through the cloud and adopting infrastructure globally."
While countries are already focused on AI sovereignty, more states may begin assessing data sovereignty, according to Amato-Cravero. “It's great if you have all the technological and infrastructure solutions, but if your law and regulation don't accommodate that you can have the data, then even if businesses are rooted in your market, they will be reliant on data from a different jurisdiction."
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.
[View Source]
