Bill Gates has accused the Trump administration of obstructing efforts to guard against AI risks, warning the technology could be exploited to trigger a pandemic.
Speaking Wednesday at an AI futures summit, the Microsoft co-founder said AI models display "astonishing" capabilities, meaning their potential to benefit humanity and their potential to cause harm cannot be overstated.
He described malicious actors using AI tools to create biological weapons, or to attack cyber vulnerabilities across society, as the most pressing risk today.
Particularly alarming, he noted, is that some so-called "open" models are freely available to developers, and the safety guardrails and monitoring mechanisms on such models can be stripped away.
The tech billionaire warned that an AI-driven biosecurity incident could have a "scale of destruction potentially far greater than a pandemic."
Despite these risks, Gates said global efforts to regulate and monitor AI and its impacts are stalled because of White House inaction.
"If the US isn't involved, the value of the effort is very limited. So we're stuck... the only question is how many bad events have to happen before they change their position."
Gates is one of the tech leaders who has repeatedly issued strong warnings. He cautions that the current development path of AI could lead to the technology being misused, or slipping beyond human control, with catastrophic consequences.
The philanthropist has previously warned that AI could cause a billion deaths and massively disrupt job markets.
In recent months, concerns about the safety risks of this emerging technology have continued to escalate. Last week, Anthropic's IPO prospectus warned that AI could pose "catastrophic or even existential risks" to humanity.
Gates praised the frontier AI lab for being candid about AI risks: "They've been very candid, and this is another example."
Despite the astonishing pace of AI iteration, President Trump has rejected calls to strengthen government regulation of AI, dismissing attempts to advance G20 AI guidelines as a "globalist scheme" and preferring self-regulation by the tech industry.
Gates believes industry self-regulation will not work: "Once multiple companies are involved, the responsibility for solving this problem can only rise to the government level."
He called on governments to set benchmarks for safety guardrails and monitoring mechanisms built into AI models, to lead global action to identify cyber vulnerabilities, and to build a large-scale monitoring system covering air, wastewater and blood to detect potential AI-induced infections as early as possible.
Gates received a pre-release version of OpenAI's ChatGPT back in 2022. He revealed that OpenAI and Anthropic will likely launch new models this year, with "two major releases" still to come in 2027.
He added that with the rollout of this new batch of models, "the models themselves going out of control" will become a serious problem in the near future.
Gates argues that governments need to further improve safety assurance mechanisms and conduct third-party audits of new models.
He also called for accelerating the pursuit of AI's positive value, such as drug development and shortening hospital waiting times: "These benefits won't happen on their own."
Gates also expressed support for taxing billionaires and ultra-high-income groups. Against the backdrop of AI profoundly reshaping society, this stance amounts to support for California's "wealth tax" ballot initiative.
According to preliminary vote counts from Tuesday's US midterm elections, the initiative is likely to be rejected.
He affirmed the philanthropic commitments of the co-founders of OpenAI and Anthropic, and said he is in talks and working with them to "ensure these funds play a huge role in global health and many other areas."
Comments