A brief statement signed by hundreds of AI scientists, academics, technology executives and public figures calls on the world to treat the risk of AI-driven extinction as a global priority. The signatories include OpenAI CEO Sam Altman, DeepMind CEO Demis Hassabis and computer scientist Geoffrey Hinton. The statement was hosted by the Center for AI Safety, a privately funded nonprofit based in San Francisco.
Its message is concise: mitigating extinction risk from AI should rank alongside other society-wide dangers such as pandemics and nuclear war. The statement has renewed debate over where policymakers should focus their attention as AI systems become more widely used.
A short statement in a crowded debate
The Center for AI Safety says the statement was kept brief so that discussion of severe risks from advanced AI would not be lost among other urgent concerns. Its director, Dan Hendrycks, has also said those other concerns deserve attention, naming systemic bias, misinformation, malicious use, cyberattacks and weaponization.
The article notes that warnings about future, highly capable AI had already received significant public attention in recent months. Hinton had warned of an existential threat from AI taking control, while Altman had called for regulation to prevent AI from destroying humanity.
An earlier open letter signed by Elon Musk and others called for a six-month pause on developing AI models more powerful than OpenAI’s GPT-4. It argued that time was needed to establish shared safety protocols for systems that their creators might not be able to understand, predict or control reliably.
The newer statement takes a different form: it urges attention to risk mitigation rather than calling for a development pause. Together, the interventions illustrate how concerns about powerful future systems have become a visible part of the policy debate.
Keeping present-day harms in view
The article argues that attention to hypothetical future dangers can come at the expense of scrutiny of problems associated with current AI tools. It points to the use of copyrighted material in training without permission, consent or payment, and the scraping of personal data online without adequate regard for privacy.
Other concerns raised include limited transparency about training data, misinformation produced by flawed systems, bias that can lead to automated discrimination, AI-generated spam and the energy required to train these systems. These issues are distinct from extinction risk, but the article presents them as part of the same policy landscape.
That distinction matters for public debate. A focus on catastrophic scenarios may make for dramatic headlines, while data practices, errors and discriminatory outcomes can be harder to examine. The article suggests policymakers and journalists should give those current effects sustained attention too.
Who shapes the rules?
The article also questions how calls to regulate advanced AI could affect competition and market power. It cites Jenna Burrell, director of research at Data & Society, who argued in a Columbia Journalism Review article that coverage should move away from speculative questions about AI sentience and examine how AI may concentrate wealth and power.
From that perspective, the debate is not only about what risks AI could pose in the future. It is also about who develops the systems, whose data they use, and who has influence over the safeguards that governments may introduce.
The article points to OpenAI’s involvement in efforts to shape democratic processes for steering AI, as well as lobbying aimed at regulators. It describes the company as seeking a role in discussions about future guardrails while its tools are being scrutinized in policy debates. These details inform the article’s argument that proposals focused on distant risks may also intersect with commercial interests and competition concerns.
The role and funding of the Center for AI Safety
The Center for AI Safety says its mission is to reduce society-wide risks from AI through research, field-building and policy advocacy. Its website describes it as a nonprofit supported entirely by private contributions and says donors do not determine its policies or research direction.
The article notes that public information about the organization and its financial backing is limited, and that the publication had contacted the center with questions. The organization’s stated position is that multiple risks can be addressed at once: it would be reckless to focus only on present harms, but also reckless to ignore them.
That argument frames the central challenge raised by the statement. Policymakers face pressure to consider severe future risks while also dealing with harms already linked to AI systems. The question is how to keep both in view, and how to ensure that the people and organizations shaping the rules are accountable for the choices they make.