Open-source AI models get a booster shot on US restrictions on frontier technology

The growing interest in open models also comes as enterprises seek to reduce AI costs.

The US government’s decision to restrict access to some of the world’s most advanced artificial intelligence models has accelerated interest in open-source AI, particularly Chinese models, as businesses seek alternatives that are less vulnerable to government controls, AFP reported.The restrictions on Anthropic’s Mythos 5 and Fable 5 models, followed by tighter oversight of OpenAI’s latest GPT-5.6 release, have reignited debate over the merits of open versus closed AI systems.Unlike closed models such as OpenAI’s ChatGPT and Anthropic’s Claude, where companies control access through their own platforms, open-source or open-weight models allow developers and enterprises to download, modify, and deploy them independently.Last month, the Trump administration directed Anthropic to block non-Americans from accessing its most advanced models.

Rather than implementing user screening, the company withdrew the models entirely.

OpenAI subsequently agreed to release GPT-5.6 only through a government-approved preview programme for selected customers.The episode has highlighted the risks businesses face when relying exclusively on proprietary AI systems.“If everything you need to do has to be on a specific frontier model, that makes whatever you’re building a whole lot less reliable when access is suddenly withdrawn,” said Oren Michels, co-founder and chief executive of Barndoor AI.Haitham Mengad, co-founder of AI music startup Stems Labs, said the removal of Anthropic’s Fable model demonstrated how dependent some developers had become on closed AI platforms.“Fable has been a game-changing model for me.

Honestly, when they took it off, it was the first time that I realised… it’s almost like a drug,” Mengad said, adding that the episode strengthened the case for open-source alternatives.Cost effectiveness of open modelsThe growing interest in open models also comes as enterprises seek to reduce AI costs.

Around the same time as the US restrictions, Chinese AI company Zhipu AI released its open-weight GLM-5.2 model, which AI analyst Andrew Curran said performed close to leading models from Anthropic and OpenAI while allowing enterprises to run it on their own infrastructure.According to data from AI routing platform OpenRouter, the combined usage share of Google, Anthropic and OpenAI models fell from 55 per cent in January to 33 per cent in June, while China’s open-source DeepSeek emerged as the platform’s most widely used model.“You want to be as flexible as you can be.

Maybe a year and a half ago, some large company might say we bought Anthropic, or we bought OpenAI, and now no one, no one buys only one,” said Michels.

The shift is also changing perceptions of Chinese AI.

Mengad said he did not believe open Chinese models posed significant security risks because, once downloaded and run locally, developers no longer have access to enterprise data or control over how the models are used.However, some experts believe government scrutiny could eventually extend to open models as they become more capable.“If Mythos-level models are considered risky, China will also not want them to be open,” said Ethan Mollick, a professor at the University of Pennsylvania, suggesting governments may increasingly seek tighter control over the most advanced AI systems regardless of where they are developed.