OpenAI CEO raises concerns about AI and calls for regulation in Congress testimony

Sam Altman’s plan for regulating AI involves creating a new government agency to license and oversee large AI models, establishing safety standards and conducting independent audits of AI.

Sam Altman hinted at OpenAI's Project Strawberry, aiming to enhance AI's reasoning abilities.

During a Congress panel hearing, Sam Altman, the CEO of OpenAI, expressed his apprehension about the potential harm caused by AI. Altman emphasised the need for regulation and clear guidelines to address these risks and ensure the responsible development and deployment of AI technology.

The hearing included testimonies from Sam Altman and witness experts and covered a range of crucial topics, including election interference, safety measures for children, disinformation, data transparency, liability frameworks, international standards, and the impact of AI on various industries.

Altman outlined his plan for regulating AI, proposing the formation of a new government agency responsible for licensing large AI models, with the authority to revoke licences from companies that fail to meet government standards. He emphasised the importance of establishing safety standards for AI models, including evaluations of their dangerous capabilities. Altman also called for independent audits conducted by external experts to ensure accountability and performance evaluation of AI models. Altman advocated for international standards in AI regulation, drawing parallels to the International Atomic Energy Agency’s role in setting global rules for nuclear weapons.

When questioned about the potential use of AI to interfere with election integrity, Altman expressed his concerns and emphasised the necessity of regulations and guidelines to address these risks, ensuring transparency and disclosure from companies providing AI models. He called for regulation, corporate policies, and public education to address the ways political entities could use AI to fine-tune their strategies in elections. Altman highlighted the safety measures specifically aimed at protecting children, assuring that they have been implemented for ChatGPT.

In terms of the liability of intermediaries, Altman shared his viewpoint that a new framework should be established specifically for generative AI tools like ChatGPT. He emphasised the significance of determining an appropriate liability framework for such technologies, stating that it is a crucial question that requires careful consideration.

The hearing also addressed concerns about the music industry and content creators. Altman emphasised the importance of content owners receiving significant benefits and having a say when their works are used to train AI models.

Senator Richard Blumenthal, Chairman of the Senate Subcommittee on Privacy, spoke about Sam Altman’s testimony after the hearing and emphasised that Congress should not be the sole regulator of AI. Blumenthal stressed the need for a broader regulatory agency and underscored the importance of providing resources and expertise to the new regulatory agency to prevent private companies from surpassing government oversight. He pointed out that the FTC lacks the capability to effectively regulate AI and highlighted additional issues, including monopolisation and national security threats, that were not addressed during the hearing. He called for cautious decision-making and advocated for an approach that avoids causing harm, ensuring the implementation of effective and enforceable rules.