Should AI development be paused or restricted?

Should we hit pause on AI? I'll map the case for and against restricting development.

There is no consensus that AI development should be fully paused, but there is broad debate over whether it should be restricted and steered through regulation. The discussion falls into three positions:

1. The Call for a Pause

In March 2023, the Future of Life Institute published an open letter calling for "all AI labs to immediately pause for at least 6 months the training of AI systems more powerful than GPT-4", citing risks such as AI-generated propaganda, extreme automation of jobs, human obsolescence, and loss of control.

The letter received more than 30,000 signatures, including academic AI researchers and industry CEOs such as Yoshua Bengio, Stuart Russell, Elon Musk, Steve Wozniak and Yuval Noah Harari.

As reported at the time:

> Recent months have seen AI labs locked in an out-of-control race to develop and deploy ever more powerful digital minds that no one – not even their creators – can understand, predict, or reliably control.

The letter warned that AI systems with "human-competitive intelligence can pose profound risks to society and humanity" — from flooding the internet with disinformation and automating away jobs to more catastrophic future risks.

The specific demand was: "We call on all AI labs to immediately pause for at least 6 months the training of AI systems more powerful than GPT-4", and that "If such a pause cannot be enacted quickly, governments should step in and institute a moratorium".

Proponents clarified the request was limited - "This does not imply a pause or ban on all AI research and development or the use of systems that have already been placed in the market".

More recently, Anthropic CEO Dario Amodei laid out a case for why AI development should be decelerated so that adequate guardrails can be deployed, and former OpenAI safety researchers have warned about existential risks and loss of control.

Arguments for pausing or restricting:

2. Arguments Against a Full Pause

Opponents argue a blanket pause is impractical and counterproductive:

Alternative proposals from this camp emphasize:
> Instead of pausing AI development, we should encourage more transparency and accountability among AI labs - by sharing data, code, and methods, publishing peer-reviewed papers, and engaging with external reviewers
and establishing clear and enforceable standards and regulations, such as defining ethical principles and creating legal frameworks for oversight.

3. What Regulation Actually Does: Restriction, Not Pause

No major economy has implemented a full pause. Instead, the two leading models are risk-based restriction:

European Union - EU AI Act: Enacted in 2024, it represents the world's first holistic framework, categorizing AI by risk levels and banning high-harm applications like social scoring. It establishes a detailed, legally binding, risk-based governance model aimed at ensuring AI operates in accordance with human dignity, democracy, and the rule of law. It classifies systems by risk and prohibits certain applications outright while subjecting high-risk systems to strict obligations including conformity assessments.

United States - Voluntary / Sectoral Approach: Rather than a comprehensive binding statute, the U.S. relies on a non-binding, innovation-supportive framework centered on Executive Order 14110 and the NIST AI Risk Management Framework. This reflects a preference for sector-specific regulation, agency guidance, and voluntary compliance mechanisms.

The U.S. approach explicitly rejects a licensing pause: The executive order states nothing shall be construed to authorize the creation of a mandatory governmental licensing, preclearance, or permitting requirement for development, and later orders expressly disclaim any authority to impose a mandatory licensing regime for frontier models.

Analysts note the EU follows a precautionary, risk-calibrated model while the U.S. promotes a minimalist, innovation-first approach with no equivalent risk classifications.

Conclusion: Paused vs. Restricted

In practice, the debate has shifted from "pause vs. continue" to "how to restrict" - what risk categories to enforce, whether guardrails should be binding law (EU model) or voluntary frameworks (U.S. model), and how to avoid both stifling beneficial innovation and leaving catastrophic risks unaddressed.