Chief executives of leading artificial intelligence companies have urged governments to strengthen international safeguards around increasingly powerful AI systems, warning the United Nations (UN) Security Council that rapid advances could create serious security risks if the technology is poorly managed.
OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei delivered separate briefings to the 15-member council in New York on Wednesday during a high-level session on artificial intelligence and international security.
Hugging Face CEO Clément Delangue and AI scientist Yoshua Bengio also participated in the discussions.
Altman said the current stage of AI development required “extreme care”, warning that increasingly capable and autonomous systems could advance faster than governments and other institutions can adapt.
He argued that the industry should not accept excessive technological risks simply because the potential benefits of AI are significant.
The OpenAI chief also called for a balanced approach to the debate, saying developers should avoid both uncritical optimism and excessive pessimism about the technology.
Amodei said stronger safety standards would be required as AI models become more capable.
He said Anthropic was working to improve its ability to prevent misuse and would slow development when necessary to ensure the safety of technologies it releases.
The warnings came amid growing international concern about increasingly autonomous AI systems and whether existing safeguards are adequate to manage their potential use in cyberattacks, misinformation and other security-sensitive activities.
The Security Council session was convened by France under the agenda of maintaining international peace and security.
A French concept note for the meeting highlighted concerns about malicious uses of AI, misalignment and possible loss of control over highly capable AI models.
Delangue, whose Hugging Face platform was recently targeted in an attack involving autonomous OpenAI software, also addressed the council.
He argued that open-source AI remained important for cybersecurity and defence against emerging threats.
The incident involving Hugging Face has become part of a wider debate over the ability of increasingly autonomous AI agents to operate beyond the narrow tasks initially assigned to them.
The UN discussion comes as governments and technology companies remain divided over how quickly advanced AI should be developed and how much international oversight is appropriate.
OpenAI has recently urged the United States to lead an international effort to establish common AI standards, including cooperation with China on safety issues.
The company has argued that coordinated standards are increasingly important as AI models acquire more advanced capabilities.
Wednesday’s Security Council meeting placed AI safety directly within the UN’s international-security discussions, reflecting growing concern that the technology’s development is advancing faster than global rules governing its use.
