TrendAI Modern Bank Heist: The Industrial Age of Cybercrime

AI agents, state-sponsored threat actors and cybercrime-as-a-service are outpacing defences in the financial sector, according to a new report from TrendAI.

The report, titled Modern Bank Heists 2026: The Machine-Speed War for Financial Control, surveyed CISOs across the financial sector and revealed a big shift in attacker strategy.

Adversaries are now pushing back during live incidents, with TrendAI data showing that 67% of institutions experienced such counter-incident response where bad actors attempted to undermine the work of defenders.

AI-backed attacks at machine speed

The TrendAI report revealed a stupendous rise in in AI-enabled attacks at 89% year-over-year rise, confirming that offensive operations can now run at the speed of the machine.

Campaigns that once needed skilled operators, including phishing and fraud, are increasingly run in the background without direct human input, aided by AI tools.

"Cybercrime has entered its industrial age," says Bharat Mistry, Field Chief Technology Officer at TrendAI.

"Criminal organisations are chaining together AI agents that can conduct reconnaissance, launch phishing campaigns, evade detection and exploit vulnerabilities with minimal human intervention.

"Financial institutions are no longer facing isolated attacks, they are confronting highly automated adversaries operating at machine speed."

Destructive attacks and stolen intelligence

Of those surveyed, 41% suffered a cyberattack that was destructive in nature.

TrendAI describes this as indicative of a "deliberate shift from exfiltration toward damage".

The report also recorded a 55% increase in API-based attacks over the same period.

Criminals are also targeting non-public market intelligence, with 46% of respondents reporting attempts to steal investment strategies – an area TrendAI says is now pursued as aggressively as funds themselves.

Account takeover was named the most pressing threat by 37% of respondents.

Business email compromise (BEC) and reverse business email compromise (RBEC) schemes, both enhanced by deepfakes, followed at 28%.

Nearly four in 10 (37%) of respondents confirmed instances of "island hopping", where attackers compromise an organisation's infrastructure and then use it as a launch point against customers.

All of this comes at a time when 54% of organisations report no increase in security budget, a situation TrendAI describes as leaving security leadership "structurally subordinated".

Steganography hides attack traffic

To conceal attack infrastructure, cybercriminals are embedding commands within image pixels that malware can retrieve later.

One example cited by TrendAI is the Daserf backdoor, which uses steganographic algorithms such as RC4 alongside base64 encoding.

This method allows Daserf to establish covert command-and-control channels that can pass beneath traditional traffic inspection.

Nation-state actors, including Pawn Storm, use steganography in combination with legitimate cloud services to avoid endpoint monitoring altogether.

Cybercriminals are also known to exploit compromised cloud storage buckets to distribute malicious code hidden inside images.

A method that did not exist two years ago, according to TrendAI, is invisible prompt injection, where malicious instructions embedded in images are silently processed and acted on by AI systems.

The RAT plague threatening finances

TrendAI identifies five remote access trojans (RATs) as posing a particular threat to financial institutions, each confirmed to have been used against banks or cryptocurrency exchanges:

• Remcos: once marketed as a legitimate tool, this is now used as a real-time surveillance platform capable of live webcam streaming and instant keystroke transmission.
• AsyncRAT: a free open-source RAT described by TrendAI as the most prolific by volume, was linked to the advanced persistent threat (APT) group Winnti's GodRAT campaign distributed through Skype.
• XenoRAT: forked by North Korea's Kimsuky APT group into a variant called MoonPeak, deployed against South Korean financial and government entities with an impact running into billions of dollars.
• BananaRAT: combines screen streaming, overlay injection, QR code-based Pix transaction manipulation and continuous keylogging in a single platform, operated by the Brazilian group SHADOW-WATER-063 against 16 financial institutions and cryptocurrency exchanges in Brazil.
• XWorm: offered as malware-as-a-service with a lifetime licence priced at US$500, described by TrendAI as covering every major financial attack vector from a single implant.

"It is the only tool in this class that covers every major financial attack vector from a single implant," TrendAI says of XWorm.

The report concludes that financial institutions need to move beyond reactive cybersecurity and adopt a proactive intrusion suppression strategy.

The company recommends combining AI-powered detection with proactive threat hunting, virtual patching and managed detection and response to strengthen resilience against increasingly autonomous attacks.

TrendAI also advises building AI-enabled security operations capable of responding at machine speed, strengthening protection against prompt injection, deepfake-enabled fraud and business email compromise and elevating CISOs into independent executive roles with direct responsibility for cyber resilience.