Technology and Power — Essay Series
How emerging technologies reshape economic power, governance and global competition.
When AI Innovation Moves Beyond Intellectual Property: Safety, Secrecy and Power
The core issue
AI is moving innovation outside the intellectual property system into secrecy and controlled infrastructures.
This is not simply a technical legal development — it is a shift in power.
Power implications
- Disclosure collapsing
- Sovereignty becoming control
- Openness becoming strategy
Introduction: AI and the changing structure of innovation
The rapid acceleration of artificial intelligence is transforming not only technological capabilities but also the structures through which innovation is governed and diffused. While public debate has focused heavily on safety, regulation and competitiveness, less attention has been paid to a deeper structural shift: the changing relationship between AI development and intellectual property systems.
For more than a century, intellectual property has operated through a foundational bargain. Innovators receive protection in exchange for disclosure, enabling knowledge diffusion, cumulative innovation and wider societal benefit. This system has underpinned modern technological progress and shaped global innovation governance.
Today, however, leading-edge AI development is increasingly occurring outside this framework. Proprietary datasets, closed models, compute-intensive infrastructures and trade secrecy are becoming the dominant foundations of competitive advantage.
If AI innovation moves beyond the IP system, intellectual property does not become obsolete — it becomes irrelevant.
These reflections draw on recent parliamentary discussions, including a UK All-Party Parliamentary Group (APPG) roundtable convened to examine the implications of artificial intelligence for intellectual property and technological power.
This shift has profound implications not only for innovation incentives, but also for safety, competition, and the distribution of technological power.
Why this matters now
If innovation increasingly takes place within closed ecosystems governed by secrecy and infrastructure control, traditional tools of oversight and diffusion weaken. Knowledge flows become restricted, competitive advantages concentrate, and the capacity of states and institutions to shape technological trajectories may diminish.
The question is therefore not simply how to regulate AI safely, but how to understand the changing relationship between innovation, governance and power in an era where the traditional IP bargain is eroding.
The erosion of the disclosure bargain
Patent systems historically incentivised innovators to disclose technical knowledge in return for time-limited exclusivity. This disclosure supported cumulative innovation and allowed technological capabilities to spread across sectors and borders.
In AI development, however, many of the most valuable assets are not easily captured through traditional patent mechanisms. Training datasets, model architectures, parameter weights and optimisation techniques often remain protected as trade secrets or embedded within proprietary infrastructures. Even where patents are used, they may disclose relatively little of practical value compared with the strategic importance of data and compute resources.
As a result, the traditional disclosure bargain is weakening. Firms frequently derive greater advantage from maintaining secrecy and control than from participating in formal intellectual property systems. Over time, this may reduce the volume of publicly available technical knowledge and challenge the foundations of innovation diffusion.
Trade secrecy, infrastructure and concentration
The increasing reliance on trade secrecy and infrastructure control is altering competitive dynamics. Advanced AI development requires access to vast datasets, specialised talent and substantial computational resources. These requirements favour already dominant firms and create structural barriers to entry.
When innovation is primarily protected through secrecy rather than formal IP rights, competitive advantage becomes embedded in organisational scale and ecosystem control. This concentration extends beyond markets into geopolitical and strategic domains, where control over AI capabilities may translate into economic and political influence.
Safety considerations are also affected. Systems developed within highly closed environments limit opportunities for external scrutiny, independent verification and shared learning. In such contexts, safety becomes closely tied to internal corporate governance rather than broader institutional oversight.
AI innovation paths
The emerging AI landscape can be understood through three distinct innovation pathways:
AI innovation paths
- Patent → disclosure → diffusion
- Trade secret → control → concentration
- Open model → access → ecosystem growth
These pathways are not mutually exclusive, but their relative prominence will shape future innovation systems. A landscape dominated by secrecy and control risks reinforcing concentration and asymmetries of knowledge. One that incorporates structured openness and shared infrastructures may support broader participation and distributed oversight.
Openness as strategic infrastructure
Openness in AI is often framed as a normative or ethical choice. Increasingly, however, it is becoming a strategic one. Open models, shared datasets and collaborative development environments can function as counterweights to excessive concentration of technological capability.
Such ecosystems enable broader participation in innovation, facilitate transparency and allow safety practices to evolve collectively. At the same time, they raise complex questions regarding security, misuse and sustainable economic models. The challenge is not simply to promote openness, but to design forms of openness that enhance both innovation and safety.
For governments and public institutions, openness may also serve as a form of technological sovereignty. Investment in shared infrastructures and collaborative research environments can reduce dependence on a small number of dominant providers while fostering domestic capability and resilience.
If AI innovation moves beyond the IP system, intellectual property does not become obsolete — it becomes irrelevant.
Implications for policy and governance
The movement of AI innovation outside traditional IP frameworks presents a strategic challenge for policymakers. Conventional intellectual property tools may have limited influence over developments occurring within closed, data-driven and infrastructure-based environments.
Ensuring safety and broad societal benefit therefore requires a wider governance lens. Competition policy, data governance, public investment in open infrastructures and new forms of international cooperation will play increasingly central roles. Effective oversight must extend beyond formal rights into the governance of platforms, datasets and computational resources.
Industry actors also face strategic choices. Short-term advantage gained through secrecy and control may contribute to systemic risk and long-term instability. Conversely, structured openness and responsible disclosure can help build resilient and trusted innovation ecosystems.
What must be avoided
Three pitfalls are particularly significant:
- Equating secrecy with safety. Closed systems may reduce transparency and external scrutiny.
- Assuming existing IP frameworks are sufficient. Traditional tools may not address infrastructure-based innovation.
- Neglecting ecosystem design. Safety and innovation depend on collaborative structures as much as regulation.
Avoiding these pitfalls requires coordinated action across public and private sectors, supported by international dialogue and strategic foresight.
Conclusion: technology and power at an inflection point
Artificial intelligence is reshaping the relationship between innovation, intellectual property and power. The central question is no longer whether AI can be governed within existing IP frameworks, but whether innovation will continue to evolve beyond them.
If it does, the future of innovation governance will depend less on formal legal protections and more on how societies manage control over data, compute and collaborative ecosystems. The choices made now will determine whether AI reinforces concentration and opacity or supports a more open, balanced and resilient technological order.
Technology and Power — Essay Series
This series explores how emerging technologies reshape economic structures, governance systems and global power relations.
Professor Birgitte Andersen is Professor of the Economics and Management of Innovation and leads research on the political economy of emerging technologies.
