The chair of the Australian Securities Investments Commission (ASIC), Joe Longo, has identified gaps in legislation relating to preventing and responding to harms caused by machine learning and automated decision-making. Existing laws are insufficient to prevent damage caused by opaque AI systems, and regulations tailored to crimes committed through algorithms or AI would be more effective in preventing them. While current laws are sufficient to punish bad actors, their ability to avoid harm may not be. Regulations surrounding responsible outsourcing may apply, but they may not prevent injury. There is a need for transparency and oversight to avoid unfair practices, whether accidental or intended, but the current regulatory framework may not ensure this. Laws protecting consumers against AI-facilitated harms must address the non-transparency when using AI, unintentional bias, the difficulty of appealing an automated decision and establishing liability for damages.
Source: IT News
Leave a Reply