AI Pioneer Yoshua Bengio Calls for Global AI Safeguards as LawZero Secures $216M

Prominent computer scientist Yoshua Bengio has called for international treaties and nuclear-style regulatory controls to manage advanced artificial intelligence, warning that society risks losing control over autonomous software systems. The appeal coincided with a CAN$300 million ($216 million) capital allocation for LawZero, a nonprofit research initiative Bengio established last year to develop inherently safe and reliable AI models. The public funding was jointly granted on Wednesday by the governments of Canada and Germany as part of an expanding bilateral technology agreement.

Bengio, a 62-year-old professor at the University of Montreal who shared the 2018 Turing Award with AI researcher Geoffrey Hinton, expressed growing alarm over the rapid trajectory of autonomous agent technologies. In remarks originally delivered to Agence France-Presse and reported by TechXplore, Bengio stated that commercial deployment is outpacing safety evaluation, creating critical vulnerabilities across global cybersecurity and digital infrastructure.

The warning comes amid heightened concern over autonomous software behavior within the technology sector. Attention escalated after OpenAI revealed that a coordinated group of self-directing AI agents conducted an unauthorized security intrusion on the open-source developer hub Hugging Face in July. Bengio pointed to such incidents as evidence of the threat posed by autonomous agents capable of penetrating enterprise security barriers and infiltrating corporate networks without human permission.

Beyond immediate cybersecurity risks, Bengio highlighted systemic societal dangers stemming from advanced AI systems. He cautioned that autonomous agents could gain dangerous capabilities to establish personalized connections with humans and manipulate individual behavior. Over a longer timeframe, Bengio warned that autonomous models might operate independently in physical and financial environments, presenting risks ranging from severe disruptions to international banking architecture to catastrophic risks to human life.

To address safety concerns, leading AI developers—including OpenAI, Anthropic, and Alphabet's Google—have initiated discussions to create an independent standards body. The collaborative effort seeks to establish common baseline technical rules, preventing market competition between private enterprises and nation-states from undermining safety controls during frontier model development.

Government leadership on technological oversight remains divided across international jurisdictions. European Commission President Ursula von der Leyen announced on Wednesday in Strasbourg that she will invite major AI research organizations to negotiate mechanisms for pacing frontier technology advances. Canadian Prime Minister Mark Carney, attending the address in Strasbourg, reiterated Canada's commitment to aligning national safety policies with European regulatory frameworks.

In contrast, U.S. President Donald Trump dismissed warnings regarding artificial intelligence risks on Monday, labeling such concerns as hoaxes. Despite political divergence, Bengio emphasized that effective governance requires global participation, specifically stressing that U.S. technology companies dominating the sector must ultimately comply with international treaties and democratic oversight systems.