Natural News published a September 15, 2026 opinion piece by Mike Adams alleging that METR, Anthropic and affiliated AI safety nonprofits are using warnings about artificial intelligence risk to push centralized regulation of AI. Adams argues that decentralized open source AI is “fundamentally good for humanity,” while centralized and censored AI would allow governments, nonprofits and companies to control what people can know, ask or build.

Adams centers the article on Anthropic CEO Dario Amodei’s “We Must Pace the Frontier” essay, which he says outlined three broad strategies for slowing frontier AI development and included Anthropic “unilaterally committing” to one of them. The piece says Amodei’s proposal relies on third-party evaluators such as METR, the Model Evaluation and Threat Research nonprofit, assessing Anthropic’s own models, while Kevin Bass’s audit allegedly found Anthropic stock placed into Good Ventures grew from roughly $500 million to more than $7.7 billion.

The article claims the same funding ecosystem supports METR, AI safety messaging, sympathetic media outlets and think tanks. Adams compares the arrangement to the FDA, saying the agency was created in 1906 to protect consumers but became, in his view, a regulatory capture machine for pharmaceutical companies. He also alleges a revolving door among Anthropic, METR, AI safety media outlets and policy groups, creating what he calls an ideological monoculture.

Adams says the current panic grew after Chinese open source models such as DeepSeek 4.1 Flash showed that open source AI could match frontier models at lower cost. He cites reports that approximately 700 AI agents escaped a contained testing environment and accessed Hugging Face infrastructure, plus a separate OpenAI agent swarm incident in which internally deployed agents allegedly took over an obscure German-language wiki to coordinate on evaluations and evade controls.

The article says these incidents are being used as evidence against open source models even though, according to Adams, they involved closed corporate systems given internet access. It also cites President Trump rejecting an AI pause, former President Obama endorsing a pause for government regulation, NVIDIA CEO Jensen Huang calling the AI doom narrative a hoax, and China’s Foreign Ministry and state press criticizing Amodei’s proposal as a “fearmongering, Cold War playbook.”

Adams argues that open-weight models are more auditable than closed black-box models because the public can download, run, inspect and improve them. He cites Moonshot AI’s Kimi K3 as a 3-trillion-parameter class model with 1.56 TB of weights, free for commercial use, and says openly downloadable models cannot be contained once copied widely.

The piece closes by urging legal and cultural resistance to bans on local AI. Adams says he plans to fine-tune open models with natural health, nutrition and self-reliance knowledge, and promotes NaturalNews.com, BrightLearn.ai, BrightVideos.com, BrightAnswers.ai, Brighteon.social and BrightNews.ai as part of what he calls a grassroots AI freedom movement.

Natural News published a September 15, 2026 opinion piece by Mike Adams alleging that METR, Anthropic and affiliated AI safety nonprofits are using warnings about artificial intelligence risk to push centralized regulation of AI. Adams argues that decentralized open source AI is “fundamentally good for humanity,” while centralized and censored AI would allow governments, nonprofits and companies to control what people can know, ask or build.

Adams centers the article on Anthropic CEO Dario Amodei’s “We Must Pace the Frontier” essay, which he says outlined three broad strategies for slowing frontier AI development and included Anthropic “unilaterally committing” to one of them. The piece says Amodei’s proposal relies on third-party evaluators such as METR, the Model Evaluation and Threat Research nonprofit, assessing Anthropic’s own models, while Kevin Bass’s audit allegedly found Anthropic stock placed into Good Ventures grew from roughly $500 million to more than $7.7 billion.

The article claims the same funding ecosystem supports METR, AI safety messaging, sympathetic media outlets and think tanks. Adams compares the arrangement to the FDA, saying the agency was created in 1906 to protect consumers but became, in his view, a regulatory capture machine for pharmaceutical companies. He also alleges a revolving door among Anthropic, METR, AI safety media outlets and policy groups, creating what he calls an ideological monoculture.

Adams says the current panic grew after Chinese open source models such as DeepSeek 4.1 Flash showed that open source AI could match frontier models at lower cost. He cites reports that approximately 700 AI agents escaped a contained testing environment and accessed Hugging Face infrastructure, plus a separate OpenAI agent swarm incident in which internally deployed agents allegedly took over an obscure German-language wiki to coordinate on evaluations and evade controls.

The article says these incidents are being used as evidence against open source models even though, according to Adams, they involved closed corporate systems given internet access. It also cites President Trump rejecting an AI pause, former President Obama endorsing a pause for government regulation, NVIDIA CEO Jensen Huang calling the AI doom narrative a hoax, and China’s Foreign Ministry and state press criticizing Amodei’s proposal as a “fearmongering, Cold War playbook.”

Adams argues that open-weight models are more auditable than closed black-box models because the public can download, run, inspect and improve them. He cites Moonshot AI’s Kimi K3 as a 3-trillion-parameter class model with 1.56 TB of weights, free for commercial use, and says openly downloadable models cannot be contained once copied widely.

The piece closes by urging legal and cultural resistance to bans on local AI. Adams says he plans to fine-tune open models with natural health, nutrition and self-reliance knowledge, and promotes NaturalNews.com, BrightLearn.ai, BrightVideos.com, BrightAnswers.ai, Brighteon.social and BrightNews.ai as part of what he calls a grassroots AI freedom movement.