OpenAI CEO Delays IPO Amid Growing AI Safety Concerns

OpenAI CEO Sam Altman announced that the company will not pursue an Initial Public Offering (IPO) this year, citing increasing concerns about the safety of artificial intelligence. Altman stated in an interview with Fortune that an IPO at this stage would be "unwise," potentially pushing back the highly anticipated offering until at least 2027. This contrasts with a statement made by OpenAI CFO Sarah Friar just last month, who suggested an IPO in 2027 or sooner if business growth continued.

These concerns are echoed by Anthropic CEO Dario Amodei, who published an article calling for AI companies to slow down the pace of advanced model upgrades. Amodei proposed a three-step plan: granting external evaluators broader access to test safety measures and report anomalies, establishing common safety standards among AI companies in democratic nations, and fostering cooperation between democratic and authoritarian governments on AI safety. He clarified that "slowing down" means allowing sufficient time for model testing, alignment with human values, and safety assurance, not halting development.

Altman and Elon Musk have both expressed support for Amodei's call to slow down advanced AI development, despite their business rivalry. OpenAI's Chief Scientist, Jakub Pachocki, also warned that no AI company has adequately solved alignment and monitoring issues, making rapid expansion difficult to justify. He hopes a voluntary slowdown will become the norm until shared safety standards are established.

The safety concerns extend beyond the industry, with bipartisan pressure growing in Washington for AI regulation, testing, and safeguards, particularly after cyberattacks potentially executed without direct human control. Concurrently, communities across the U.S. are opposing the construction of AI data centers due to their high electricity and water consumption.

Amodei believes AI can dramatically improve quality of life but suggests a one or two-year slowdown for enhancing safety and oversight mechanisms could significantly reduce the risk of catastrophic events. However, he cautioned that excessive delays could benefit authoritarian states. In related news, Tel-Hai College in Israel is launching a Faculty of AI Systems Engineering.