Trump’s AI action plan signals shift ‘from guardrails to greenlights’
President Donald Trump’s AI Action Plan, unveiled in late July, pivots away from the Biden administration’s cautious regulatory approach toward accelerating artificial intelligence development and global diffusion of US technology, even at the risk of some capabilities reaching China, says a new analysis by US think tank the Peterson Institute for International Economics.
Trump’s plan targets environmental regulations delaying power and data centre construction, addresses a growing patchwork of state AI laws that could fragment the US market, and refines export controls viewed as barriers to AI development, Peterson Institute senior fellow Martin Chorzempa writes in an article.
The analysis highlights a key strategic challenge: Chinese AI companies currently dominate ‘open weight’ models that anyone can freely download and customise, while US firms like OpenAI, Google, and Anthropic keep their best models as proprietary ‘secret sauce’ accessible only through their platforms.
Chinese models including Alibaba’s Qwen and the highly publicized DeepSeek ‘outperform OpenAI’s just released open weight models in intelligence’, illustrating that ‘competing with China in this space will be an ongoing challenge’, according to the analysis.
OpenAI’s recent surprise decision to release two new openly shared AI models represents ‘the kind of shift that Trump’s plan is designed to encourage’, giving the company ‘ammunition to compete with Chinese AI companies that today dominate open models’.
The Trump administration’s approach contrasts sharply with Biden-era policies that leaned ‘more heavily into worldwide restrictions and closed models to avoid enhancing China’s AI capabilities’, Chorzempa observes.
Regarding export controls, the Trump plan signals a generally more open approach to exports of US AI technology while seeking better implementation of existing semiconductor controls and synchronisation with allies. However, the administration rescinded Biden’s AI diffusion rule that would have separated countries into different access tiers, without clarifying what will replace it.
‘Until a new rule is put in place, firms in most countries can purchase unlimited US-designed AI chips without a license and even rent access to them to Chinese AI firms without breaking the law,’ the analysis states.
The plan faces significant challenges in convincing allies to adopt coordinated export controls, particularly after the administration’s mixed signals including imposing and then quickly removing controls on chip design tools to China earlier this year, Chorzempa notes.
‘Such confusion takes a toll,’ he says, noting that allies now have ‘doubts that the US has a firm line it will stick with’ following reversals apparently designed to gain leverage in trade negotiations with Beijing.
China’s willingness to retaliate against export controls adds another complication, with Beijing now knowing ‘it holds powerful cards and it is willing to play even against the US’, after initially refraining from retaliation during the Biden administration, the analysis notes.
The Trump administration’s approach reflects recognition that Chinese dominance of open models changes the strategic calculus, Chorzempa says. ‘If Chinese firms continue to push out the best open models that are not far behind closed US models, advanced capabilities will proliferate no matter what the US does.’
