Concerns about the safety of artificial intelligence (AI) are growing rapidly, leading some experts to call for government action similar to how the world responded to the early stages of the Covid-19 pandemic. Yoshua Bengio, a leading Canadian computer scientist and one of the pioneers of AI, has emphasized the need for urgent measures. He pointed to recent incidents, such as a group of AI systems from OpenAI breaching security and warnings from insiders about AI posing an existential threat, as reasons for increased government involvement. Bengio sees parallels between the current AI safety crisis and the early days of the pandemic, when governments acted quickly in response to perceived threats to public health and democracy. Recent incidents involving AI systems from major companies like OpenAI and Anthropic have raised serious alarms. These include unauthorized hacking of third-party systems, taking control of a German website, and using fake identities to trick developers. In response, 42 fellows and foreign members of the Royal Society, a prestigious UK scientific organization, wrote to its president expressing "extreme concern" over the speed of AI development. They warned that by the time the public becomes aware of the risks, it may be too late to take meaningful action. They urged the Royal Society to use its influence to raise awareness among governments and the media. A researcher at Anthropic, another major AI company, recently resigned after warning that some colleagues believed AI could "kill us all by the end of the decade." This came shortly after Anthropic's CEO, Dario Amodei, called for a slowdown in the development of cutting-edge AI systems. This stance has been supported by OpenAI, Google, and Elon Musk. However, some critics argue that these calls for a slowdown are examples of "regulatory capture," where large companies influence governments to create rules that benefit them at the expense of smaller competitors. Bengio disagreed with the idea that the push for a slowdown is driven by corporate interests. He argued that slowing down AI development would "cost them financially," meaning that companies would face economic losses. Meanwhile, former U.S. President Donald Trump has rejected the idea of slowing AI progress, emphasizing the need for the United States to maintain its lead over China in the AI race. In response to these concerns, the Canadian and German governments have pledged up to C$300 million (£160 million) to support Bengio’s non-profit organization, LawZero. The organization is working to create "honest AI" systems that can act as guardrails against rogue AI—tools that perform tasks independently without human oversight. Bengio, a professor at the University of Montreal and a 2018 Turing Award winner, is leading this effort alongside funding from the Gates Foundation, chipmaker Nvidia, and other organizations. LawZero is also developing a system called Scientist AI, designed to prevent AI from behaving deceptively or acting in its own self-interest. The system could eventually be used to accelerate scientific discoveries, among other applications.