Gates says Trump inaction leaves world exposed to AI bioweapons

Bill Gates has accused the Trump administration of stymying attempts to protect against the risks of AI, warning that the technology could be used to cause a pandemic “significantly worse than Covid”.

Speaking at the FT’s Future of AI Summit on Wednesday, the Microsoft co-founder said that AI models’ “stunning” capabilities meant “the potential for good, the potential for harm can’t be overstated”.

Bad actors using AI tools to create bioweapons or attack cyber vulnerabilities across society posed the greatest immediate risk, he said, particularly given the ability to remove safeguards and monitoring from certain so-called “open” models that are freely accessible to developers.

The “scale of the damage” of a biosecurity incident involving AI could be “significantly worse than Covid”, the tech billionaire warned. An estimated 15mn deaths were linked to the coronavirus pandemic in 2020 and 2021.

Despite these fears, international efforts to regulate and monitor AI and its impacts were being hindered by White House inaction, Gates said.

Recommended

“There’s not much value in acting unless the US is part of that. So we’re kind of stymied . . . The only question is how many bad events need to take place before they change their posture,” he said.

Gates is among the leading tech figures to have expressed strong warnings that AI is developing in ways that could lead to it being misused or becoming beyond human control with disastrous consequences. The billionaire philanthropist has previously warned that it could cause a billion deaths and lead to widespread job disruption.

Concerns over the safety of the nascent technology have ignited in recent months, with Anthropic’s IPO prospectus last week warning of the potential “catastrophic or existential risks to humanity”. Gates praised the frontier lab’s honesty about AI dangers: “They’ve always been very open, and this is another example of that.”

Despite the blinding pace of the technology’s development, President Donald Trump has rebuffed calls for stricter government regulation of AI, dismissing an attempt to develop G20 AI guidelines as a “globalist scheme” and instead favouring self-regulation within the tech sector.

Self-regulation would not work, Gates said. “As soon as we add multiple companies, then the only way to solve this moves up to the government.”

He urged the administration to consider a benchmark for models’ internal safeguards and monitoring, and to lead a global effort to fix cyber vulnerabilities and institute “massive” monitoring efforts across “air, sewage, blood” to catch potential AI-enabled infections early.

In a sign of AI’s continuing fast-paced development, Gates — who was given pre-release access to ChatGPT by OpenAI in 2022 — said OpenAI and Anthropic would probably release new models this year, as well as “two major releases” in 2027.

The issue of “the models themselves being out of control” would become a serious concern in the near future, potentially with those upcoming models, he added. Gates argued this required further government work on safeguarding and third-party auditing of new models.

Recommended

He urged accelerated efforts to reap AI’s potential benefits, such as drug discovery and alleviating hospital waiting lists: “None of the good stuff happens naturally.”

Gates also declared his support for “tax[ing] billionaires” and the ultra-wealthy in a society uprooted by AI, in a seeming endorsement of California’s “billionaire tax”, a state ballot measure that appears on course to be defeated according to early vote counts from Tuesday’s US midterm elections.

He praised the philanthropic commitments of the OpenAI and the Anthropic co-founders, and said he was talking to them about working together “to make sure those dollars have huge impact in many areas, including global health”.