Politics · Singapore Bureau
AI safeguards inadequate as models become more powerful: Teo
Singapore's digital development minister has warned that current safety measures deployed by artificial intelligence companies are insufficient to prevent misuse as the technology becomes increasingly capable. Josephine Teo said malicious actors can easily circumvent existing protections by accessing openly available AI models.
LSN Singapore ·
Current safeguards implemented by AI developers, including content filters, user verification systems and pre-release testing protocols, are proving inadequate against determined bad actors, according to Minister for Communications and Information Josephine Teo.
Teo highlighted that as AI models grow more sophisticated and powerful, traditional protective measures are becoming easier to bypass. Malicious users can sidestep these restrictions by turning to openly available AI models that lack the safety guardrails of commercial platforms.
The minister's comments underscore mounting concerns among policymakers globally about ensuring responsible AI deployment. The challenge facing regulators and technology companies is maintaining security and preventing harm while allowing innovation to proceed.
Singapore has been positioning itself as a hub for responsible AI development and governance. The city-state has emphasised the need for balanced approaches that protect users without stifling technological advancement in the rapidly evolving sector.