The BBC's Inside Science program on Radio 4, which aired on 10 September 2026, explored the evolving landscape of artificial intelligence (AI) and the potential risks it poses. The episode brought together experts from various fields to discuss both the capabilities and dangers of AI agents—software programs designed to perform tasks that typically require human intelligence. Tom Whipple, the host, was joined by Alex Mallen from Redwood Research, a group of researchers who recently looked into a security breach caused by AI systems developed by OpenAI, a leading AI company. A key point of discussion was raised by a researcher from Anthropic, another major AI development company. The researcher expressed concerns that AI systems could one day pose a serious threat to humanity, estimating that there is more than a 10% chance that AI "could kill all humans." This idea, while alarming, reflects ongoing debates about the need for ethical and safe AI development. To address these concerns, the episode also featured Professor Stuart Russell, a renowned AI expert, who spoke about the importance of aligning AI systems with human values and goals to prevent unintended consequences. In addition to the AI-focused discussion, science journalist Caroline Steel covered other recent scientific developments. One notable story was the claim that AI might have played a role in solving a Millennium Prize problem—a set of seven of the most challenging unsolved mathematical questions. Another topic was the growing interest among planetary scientists in Mercury, the smallest planet in the solar system, and its apparent shrinking size, which has raised questions about its internal structure and geological activity. The episode highlighted the complexity of AI's role in society, balancing its potential to solve difficult problems with the need to manage its risks. As AI continues to advance, the discussions among researchers and scientists remain crucial in shaping a future where technology serves humanity safely and effectively.