A wide-ranging Treasury report says the financial industry must ensure that the new generation of bots complies with existing regulations.
It will probably end up as another well-intentioned governmental missive lost in the detritus of what is becoming the great American quadrennial democratic zigzag.
In hindsight, we might consider its release a bit of super-unfortunate timing, given that we are weeks away from Trump 2.0 and, if the President-elect’s new BFF Elon has anything to say about it, the certainty of an entirely new digital order.
Keeping Tabs
But as we wait for the new crypto-regime, and a congressional budget plan - or two - to percolate down the wire, we thought it might be a good idea to at least put down in writing some of the more cautionary messages in the report (direct link) on artificial intelligence related to financial services that was published overnight.
In it, the Treasury makes emphatically clear at the end on page 45 that financial firms must keep their AI tools, including Gen AI, well-behaved.
Immediate Action Required
«Financial firms have a legal obligation to ensure their AI usage complies with existing laws and regulations,» it said.
It added that the institutions needed to take immediate action if they noted any deficiencies.
Deepfaked Hysteria
«These actions may include updating policies and procedures, updating AI models, or switching to alternative AI providers. Additionally, as firms consider expanding AI use cases, firms should conduct similar due diligence to ensure their continued compliance with the applicable laws and regulations,» it said.
That is a bigger ask than it looks on the surface, as it includes keeping tabs on the ability of AI-enabled tools to «thwart» ID&V processes with deepfakes, an issue that finews.asia has discussed at length recently.
Stop Hallucinating
But that also means babysitting some of the newer Gen AI modulations, including hallucinations and bias, something that appears to be rather unique to them – in the non-living space as we know it of course.
«As a result, AI systems may inadvertently reinforce existing prejudices, create unfair outcomes, or lead to potential legal violations. For example, multiple respondents raised concerns regarding bias in credit underwriting and decision-making,» the report indicates.
More Models
This is not an insubstantial ask, given that the Treasury indicated that recent surveys show that 78 percent of financial institutions are implementing Gen AI for at least one use case and that 86 percent expect to see an increase in their so-called «model inventory» because of its adoption.
At this point, however, it is too soon to say what the eventual outcome will be, something the Treasury is also at pains to point out.
Gradual Integration
«Many respondents highlighted that the use of emerging AI technologies is in its early stages, only being gradually integrated in the financial services sector, and mainly beginning with business functions such as risk and compliance, client engagement, code assistance, and content extraction,» the authors indicate.
But the writing is on the wall, as payment providers use AI to analyze sales data to make customer recommendations.
Robo-advisor Investment Tips
They added that credit underwriting functions employ machine learning to look for alternative forms of data when evaluating creditworthiness while investor-related functions use it for investment and trading.
«For example, robo-advisors offer personalized investment advisory services, while AI-driven insights improve forecasting and trading process automation. AI can also inform trading strategies by identifying patterns, optimizing execution, managing portfolio workflows, and assessing risk-return tradeoffs,» they indicate.
Obsolete on Arrival
That is all well and good, but the catch is that they just must make sure that everything stays abreast of current laws.
Well - for the next four weeks at least. After that, we have to see although we thought it wise to indicate what the Treasury's standpoint once was - for posterity.
