Yoshua Bengio, Co-President of LawZero, said that the project "Mythos" demonstrates that artificial intelligence can be weaponized [1].
This warning highlights a critical shift in the AI landscape, suggesting that the technology has evolved from a productivity tool into a potential instrument of conflict. As capabilities grow, the ability to deploy these systems for harm becomes a primary security concern for global regulators and developers.
Bengio discussed these concerns during an appearance on Bloomberg Television. He said that the current state of the technology has brought the world to a stage where the risks are increasing [1].
"What Mythos is showing is that we have reached a point, and it's going to get worse, where AI can be weaponized," Bengio said [1].
The Co-President of LawZero said that the project serves as a practical illustration of these dangers. While AI has historically been viewed through the lens of economic or social disruption, the Mythos project provides a specific example of how the technology can be intentionally pivoted toward weaponization [1].
Bengio said that the trajectory of these advancements suggests the problem will not stabilize. Instead, the ability to misuse AI is expected to grow as the underlying models become more sophisticated and accessible [1].
“AI can be weaponized”
The warning from a high-level figure like Bengio suggests that the AI community is moving beyond theoretical concerns about 'existential risk' toward identifying specific projects, such as Mythos, that prove weaponization is currently possible. This shift may accelerate calls for international treaties or strict technical guardrails to prevent the deployment of offensive AI capabilities.
