Trump administration reverses AI policy after Mythos model reveals existential risks
Consensus Summary
The Trump administration has abruptly shifted its stance on AI regulation after the revelation of Anthropic’s Mythos model, which poses unprecedented risks to global infrastructure. Initially dismissing guardrails as 'woke,' the administration now seeks pre-deployment evaluations of AI models and may reintroduce Biden-era safeguards. The shift follows Trump’s earlier dismantling of regulations, including bans on 'woke' AI inputs and restrictions on Anthropic over ethical concerns. Meanwhile, China is positioning itself as a leader in AI safety standards, contrasting with the US’s commercial-driven approach. Experts like Elon Musk and Anthropic’s Dario Amodei have warned of existential threats from unchecked AI development, prompting even the Trump administration to reconsider its laissez-faire policies. The US plans to invest over $725 billion in AI this year, but the administration’s ability to balance economic growth with security risks remains uncertain.
✓ Verified by 2+ sources
Key details reported by multiple sources:
- The Trump administration reached agreements with Google, Microsoft, and xAI to allow the US Commerce Department’s Centre for AI Standards and Innovation to conduct pre-deployment evaluations of AI models before release.
- Anthropic’s Mythos model has capabilities to identify and exploit flaws in operating systems and browsers at speeds beyond human capabilities, posing threats to critical infrastructure like power, water, health, and financial systems.
- Anthropic has not released Mythos publicly but offered access to about 40 US companies to identify and remedy vulnerabilities before potential misuse.
- The Trump administration previously banned Anthropic from government contracts after it resisted using its tools for autonomous weaponry or mass surveillance, calling it a 'radical left, woke company'.
- The Trump administration initially dismantled Biden-era AI regulations, directing the Federal Trade Commission to remove barriers to AI development and prohibiting federal agencies from using models with 'woke' inputs.
- Elon Musk warned of AI’s existential risks, describing a 'Terminator scenario' where AI could 'kill us all'.
- China has emphasized AI safety, reliability, and controllability, promoting global AI governance standards, while pursuing an open-source model to compete with the US.
- The US plans to invest over $725 billion (AUD $1 trillion) in AI this year, underpinning its economy and sharemarket.
- The Trump administration is considering elevating the Centre for AI Standards and Innovation and reintroducing some Biden-era guardrails after the Mythos revelation.
Points of Difference
Details reported by only one source:
- David Sacks, the White House’s former AI 'czar', cited First Amendment concerns for opposing Democrat-led state AI regulations, stating: 'We don’t like seeing blue states trying to insert their woke ideology in AI models.'
- The article mentions a planned meeting between Trump and Xi Jinping where technology, including AI, is expected to be a major discussion topic.
Contradictions
Conflicting information between sources:
- The articles contain identical content with no contradictions between the two sources.
Source Articles
Trump goes ‘woke’ with a sudden change of mind
The AI alarm bells are ringing louder for Donald Trump, prompting an abrupt U-turn.
Trump goes ‘woke’ with a sudden change of mind
The AI alarm bells are ringing louder for Donald Trump, prompting an abrupt U-turn.