Last June, the US government signed an executive order called “Promoting Advanced Artificial Intelligence Innovation and Security.” Less than two weeks later, Anthropic decided to switch off its two most capable models, Fable 5 and Mythos 5, after the US government issued an export-control directive prohibiting any foreign entity from accessing them. Anthropic opted to suspend the LLMs entirely to guarantee compliance, which they have recently reverted.
Together, these events illustrate a significant shift. The US has spent the past year and a half loosening the rules on AI, but it is now slowly reversing that position.
For businesses in APAC, the changes matter in practical ways.The first is the impact on companies that sell into the US. The executive order relies on voluntary cooperation rather than hard mandates, asking developers to share new models with the government before releasing them, while pushing for stronger testing and cyber defence.
This is expected to affect American AI labs, but there are also questions about what it may mean for large US buyers that produce or integrate AI into their products and services. By extension, if you’re an APAC company exporting software, cloud services, or anything with AI embedded to US customers, expect to be asked harder questions about the models used, their security and how they’re tested.
Another possible impact concerns the use of US models and how regional companies may react to recent developments. The Anthropic episode shows that access to LLMs can change and affect regional businesses that rely on them. This is likely to lead companies to look for ways to minimise dependency risks and increase their resilience.
The third is what regional governments do next. Regional policymakers tend to watch Washington when making decisions, and the new scenario could nudge them in two different directions.
Some will see the US approach – risk-based, security-focused and tied to national interest – and learn from it. For APAC governments, this means resisting the urge to write one sweeping AI law and instead targeting rules at use cases that carry the most risk. It also means leaning on security testing and clear standards rather than blanket restrictions.
Others will look at the Anthropic ban and conclude that relying on models from another country is risky. In parts of APAC that are already sensitive to technological dependency, expect more talk of sovereign cloud, local hosting, trusted-partner arrangements and tougher procurement rules for government and critical infrastructure.
But a more interesting shift may be in how the region thinks about responsibility itself.
In APAC markets, where regulatory philosophies differ widely and enforcement capacity is uneven, the likely direction is not broad, prescriptive regulation, but incentive frameworks that encourage AI vendors to weigh the societal implications of the solutions they bring to market. In that approach, the most responsible providers tend to become the examples others follow. This is a healthier outcome for everyone than a patchwork of hard restrictions.
The single thread running through all of this is that, increasingly, questions about AI won’t just be limited to technical aspects such as efficiency but will start to encompass the rules that regulate it. This shift may play out unevenly across APAC, but it is one that businesses in the region cannot ignore.
(NOTE: Cynthia Lee is the APAC Vice President, Delinea)
For Editorial Inquiries Contact:
Editor Kym Bergmann at [email protected]
For Advertising Inquiries Contact:
Group Sales Director Simon Hadfield at [email protected]
