AI governance gaps leave RIAs exposed to a new wave of cyber threats

Financial advisory firms are racing to deploy artificial intelligence many are doing so without the governance structures needed to keep client data and firm operations safe.

A recent report from Fitch found that AI poses moderate disruption risk to wealth managers and retail banks and cybersecurity consultant Michael Brice, Founder and President of BW Cyber Services in White Plains, New York, told InvestmentNews that the gap between AI adoption and AI governance is the defining risk challenge facing RIAs today.

The governance gap that technology spending cannot fix

A September 2026 benchmark study from Cerulli Associates, conducted in partnership with Vista Equity Partners and surveying 68 RIA firms, found that governance structure - not assets under management or technology spend - is the strongest predictor of AI maturity.

Firms with centralized or hybrid governance models scored more than 2.5 times higher on average than those with decentralized or undefined approaches, and chief technology officers owned AI at 75% of leading firms.

Brice says the finding mirrors what he sees in the field every day.

"There is an important distinction between having an IT manager, director, or VP overseeing technology and having a formal CTO with the authority and experience to lead AI strategy across the enterprise," he said. "CTOs own AI at 75% of leading firms, while firms with centralized or hybrid governance structures scored 2.5 times higher on AI maturity."

The second factor, Brice says, is culture. "RIAs with a culture that has traditionally been very forward-leaning with IT have similarly embraced a forward-leaning role with AI," he said. "We have seen a clear divide between the many firms that are still relying on the 'check the box' AI policies that emerged two or more years ago and forward-leaning firms that have more recently created detailed AI policies that clearly define governance, oversight, and security controls at a very granular detail."

For smaller firms without formal C-suite technology leadership, the default path is often to place AI governance under the chief compliance officer - a structural arrangement Brice views as inadequate for the challenge. "AI may be embraced tactically without being strategically implemented and governed by the C-suite at an enterprise level," he said.

When AI becomes a weapon

Brice draws a deliberate parallel between today's AI security environment and the early years of the internet - a comparison that carries a sobering implication for how quickly the threat landscape could shift.

"We are still in the early stages of AI, much like the early stages of the internet," he said. "During the internet's early years, financially motivated attacks were relatively limited; most threats were disruptive viruses and worms rather than organized crime. The use of antivirus technology did not become mainstream for most personal computer users until the early 2000s."

The latency between a new technology's arrival and criminals learning to monetize it, he argues, is now compressing. "Reflecting back on this 10-to-15-year security gap, it's clear that there was a years-long learning curve for cybercriminals to understand how to monetize technology crime," he said. "We expect AI-enabled financial crime capabilities to evolve exponentially over just the next one to two years."

For now, Brice says most AI security risks he encounters are traditional threats made more scalable through AI - phishing emails with flawless grammar, automated social engineering campaigns with global reach. But he warns of an emerging and far more serious category of attack he terms a "firm-wide AI Compromise Attack."

"Instead of only IT systems being shut down and data being stolen, the firm's entire AI operations are either shut down or poisoned to the point that its AI systems no longer function, or the use of the firm's AI can no longer be trusted," he said.

Shadow AI and the threat within

One of the less-discussed vectors Brice flags is shadow AI - tools adopted by employees without formal sign-off, running quietly alongside approved systems. Most advisory firms treat this as a data-leak concern with reputational and compliance implications. Brice argues the real risk runs deeper.

"The far greater concern is that employees may introduce malicious or compromised AI tools into the firm's environment," he said.

The practical solution, he says, already exists. "There are already relatively inexpensive tools available that can detect and prevent it. The challenge is that organizations lacking strong security leadership often do not know these tools are available."

A July 2026 Investment Management Compliance Testing Survey from ACA Group, the Investment Adviser Association, and Yuter Compliance Consulting found that 85% of respondents named AI as their top compliance concern - a 28-percentage-point jump from 2025. Fifty-nine percent have stood up formal AI governance committees, and 86% maintain an inventory of approved AI tools, according to that same survey. Shadow AI undercuts both measures.

Do not wait for the SEC

A common posture among smaller advisory firms is to treat the absence of formal AI-specific regulation as permission to defer action. Brice is direct about why that reasoning is flawed.

“The existing SEC rules related to data protection and business continuity are very clear regardless of the technologies involved,” he said. “The absence of AI-specific rules does not eliminate a firm’s existing responsibility to protect sensitive information and maintain resilient operations.”

The SEC named AI training and security controls as a review area in its 2026 examination priorities. Brice expects the first wave of AI-related enforcement actions to arrive before those rules are ever formalized. “We expect to see AI-related data breaches in the near future, potentially including AI Compromise attacks, in which the SEC will make examples of the affected firms for the rest of the industry to take heed,” he said.

Advisors waiting for a formal AI rule should note that the regulator has already signaled its posture — and that exam-stage document requests for AI governance policies are now routine even without a standalone requirement in place.

The client trust imperative

Brice describes a shift that began roughly a year ago and that he views as a leading indicator of where the broader market is heading - institutional investors began requiring the RIAs managing their capital to substantially upgrade their AI policies before regulators mandated it.

"Institutional investors took proactive steps requiring RIAs who manage funds for the institutions to materially update their AI policies to extremely exacting standards mandated by the institutions," he said. "This was a shift for us, as the security compliance process usually works the other way around - the SEC mandates a requirement, and institutional investors then incorporate it into their due diligence process."

For firms that cannot respond credibly when clients or prospects ask about AI governance, the reputational exposure is specific: investor redemptions tied to a network breach, in a period when AI-related attacks are still rare enough to carry a stigma beyond ordinary cyber events.

"Since these types of attacks have not yet become mainstream, it may have more of an adverse effect on the RIA's reputation versus the more traditional cyber events such as email compromise, to which investors have become somewhat accustomed," Brice said.

His recommended first step for any advisory firm is concrete: commission an independent AI Security Controls Assessment to identify gaps in governance, access, and data protection.