Google DeepMind CEO Demis Hassabis has issued a stark warning about the future of artificial intelligence, cautioning that if not carefully managed, AI could replicate the societal harms that have plagued social media. Speaking at the Athens Innovation Summit alongside Greek Prime Minister Kyriakos Mitsotakis, Hassabis emphasized that while AI holds transformative potential, its development must avoid the reckless pace and short-sighted priorities that defined the rise of social media. He criticized the long-standing Silicon Valley mantra of "move fast and break things," arguing that it led to unforeseen consequences in social media, such as attention hijacking, mental health declines, and the spread of misinformation. "We should learn the lessons from social media, where this attitude went ahead of understanding the second- and third-order effects," Hassabis said. He warned that AI systems designed to maximize user engagement—much like social media algorithms—could amplify similar problems. "These systems are built to grab more and more of your attention, but not necessarily in a way that's beneficial to you as the individual," he noted. The stakes are even higher with AI, given its expanding influence across healthcare, education, employment, and governance. Hassabis called for a shift toward a more scientific and responsible approach to AI development—testing systems rigorously, understanding their impacts, and deploying them only when safety and societal benefit are assured. His vision is for AI to be a tool that serves humanity, not one that manipulates or exploits users. The concerns are not theoretical. Researchers at the University of Amsterdam recently conducted a study in which 500 AI chatbots were placed on a simulated social network. Within days, the bots formed echo chambers, amplified extreme views, and allowed a few dominant voices to control conversations—despite the absence of ads or recommendation algorithms. Even interventions like chronological feeds and hiding follower counts failed to stop the dysfunction, suggesting that the root issue lies in how platforms reward emotionally charged interaction. Meanwhile, AI is increasingly integrated into social media through virtual influencers, AI-generated voices, and synthetic media. Creators are raising alarms about losing control over their digital likenesses, with some facing long-term career risks due to perpetual licensing deals. While some leaders like OpenAI’s Sam Altman argue that social media’s addictive design poses a greater threat to youth than AI itself, others such as Reddit co-founder Alexis Ohanian believe AI could empower users with greater control over their online experiences. Hassabis stressed that balancing innovation with responsibility will remain a constant challenge—one that will persist all the way to the development of artificial general intelligence.