AI firms are creating an “alien intelligence” that could prove impossible to control unless appropriate safeguards are put in place, according to Cyrus Hodes, an AI safety expert affiliated with the OECD and co-founder of The Future Society. Speaking on the latest episode of the ID Talk podcast, Hodes warned of the growing threat that generative AI technologies pose to identity security, with deepfake attacks surging by over 1,400 percent in the past year.
“The real issue is blurring reality, trust, and confidence from the public into what they see,” said Hodes. He explained that modern generative AI systems can manipulate voice and facial data in real time with little latency, making it increasingly difficult to distinguish genuine communications from fabricated ones. “Bad actors are already leveraging this technology for targeted attacks, and the risks will only grow if we don’t establish proper controls,” he added.
Deepfake technologies pose a particularly serious threat to the identity industry. Hodes highlighted recent incidents where cybercriminals used AI-generated voice and video to impersonate company executives in high-stakes financial scams. “We’re seeing scenarios where a CEO’s voice is convincingly replicated, directing employees to take fraudulent actions,” he noted. “This technology in the wrong hands can undermine organizational security in unprecedented ways.”
Beyond corporate fraud, Hodes has concerns over the broader societal implications of generative AI. “Anyone can now create realistic deepfakes at home,” he said, pointing out that consumer-grade tools are becoming increasingly accessible. While innovation continues to outpace regulation, Hodes stressed the importance of public-private collaboration to address the escalating risks.
Throughout the conversation, Hodes underscored the importance of international cooperation, particularly between global AI leaders like the U.S. and China. As co-lead of the OECD’s Safety and Assurance of Generative AI (SAFE) project, he is working to develop frameworks for evaluating and mitigating AI-related risks. “This isn’t just a technical challenge—it’s a geopolitical one,” he said. “Global coordination is crucial to ensure AI development doesn’t spiral beyond human control.”
Listen to the full episode of ID Talk to hear more from Cyrus Hodes on the future of AI safety, governance challenges, and the tools being developed to combat identity-related AI threats.
–
February 20, 2025 – by the ID Tech Editorial Team
