Australia Warned of AI Collusion Risks in Power Market, AEMC Report Finds

According to a statement from the AEMC, the growing use of artificial intelligence to manage complex battery trading is creating conditions in which algorithms could independently arrive at strategies that lead to higher prices. The report notes that the risks have escalated as more data becomes available to market participants and as AI tools become increasingly sophisticated.

Algorithmic collusion occurs when AI systems, without direct communication, test and refine pricing strategies that result in synchronized price increases. Per the statement, this kind of collusion is particularly challenging for regulators to detect, and any attempts to monitor such behavior would likely be costly and resource-intensive. The report further acknowledges that retraining algorithms to avoid this issue is not practical, and that once AI systems operating as “black boxes” reach certain conclusions, it is nearly impossible to trace how those outcomes were generated.

One potential remedy, restricting the use of auto-bidding software, was described as problematic. The AEMC pointed out that such tools are essential for handling highly complex systems and, in many cases, contribute to lowering prices. Yet, even as batteries were once seen as a solution to curbing high electricity costs, many of the new large-scale batteries are being deployed by the same dominant energy players who have historically controlled the market.

Evidence of shifting market power is already visible. Data from the Australian Energy Market Operator (AEMO) shows that in the June quarter, large batteries set wholesale prices at an average of $478 per megawatt hour, compared with $245/MWh in the same period the year before. This figure was also nearly three times the price set by gas-fired generators.

Source: Renew Economy