AI-Themed Investment Scam Network Reportedly Used Keitaro Cloaking Across 15,500 Domains

Researchers reported that threat actors abused Keitaro Tracker infrastructure to cloak and route AI-themed investment scams across about 15,500 domains. The campaigns allegedly used AI-trading claims, deepfake videos or imagery, generative AI-produced copy and visuals, spoofed news pages, and fraudulent trading platforms to target potential victims while hiding scam content from scanners and reviewers.

Risk classification

• Primary risk domain: 4 Malicious actors
• Primary risk subdomain: 4.3 Fraud, scams, and targeted manipulation

The incident involves threat actors using generative AI deepfakes, images, and text to execute large-scale financial fraud and investment scams targeting individual victims.

Additional risk subdomains

• 3.1 False or misleading information: The campaign generated highly realistic fake news articles and deepfake broadcasts to mislead users into believing the scams were legitimate.

Causal factors

• Entity: Human
• Intent: Intentional
• Timing: Post-deployment

The risk was caused by human threat actors intentionally using deployed generative AI tools to create deceptive assets for financial scams.

EU AI Act risk tier

• Risk tier: 3 Limited Risk

Risk Level 3: Limited Risk. The report describes the use of 'AI-generated content such as deepfakes' and 'chatbots' (specifically, deepfake video ads impersonating news anchors and fake chatbots used as landing pages), which fall under the transparency obligations of Limited Risk systems.

AI system and alleged parties

• AI system: DALL-E, unspecified generative AI (OpenAI)
• AI purpose: Deepfake Video Generation; Social Media Content Generation
• Behaviour type: Tool
• Alleged developer: Unknown voice cloning technology developers, Unknown image generation technology developers, Unknown generative AI developers, Unknown deepfake technology developers, Unknown AI video generation technology developers
• Alleged deployer: Unknown operators of malicious Keitaro infrastructure, Unknown cryptocurrency scammers, Unknown AI-themed investment scammers, Scammers, Quantum AI scammers, Quantum AI, FaiKast
• Alleged harmed parties: social media users, News consumers, Investment scam victims, Impersonated public figures, General public, Epistemic integrity, Cryptocurrency scam victims, Cryptocurrency investors

Harm severity

Severity ratings are based on this scale.

Highest direct severity in any category: Minor. Severity is scored from Negligible to Catastrophic in each harm category, for harm the reports describe as caused directly or indirectly by the AI system.

• Physical: direct Negligible, indirect Negligible
• Infrastructure: direct Negligible, indirect Negligible
• Property: direct Negligible, indirect Negligible
• Financial: direct Negligible, indirect Minor
• Environmental: direct Negligible, indirect Negligible
• Malicious content: direct Minor, indirect Minor
• Differential treatment: direct Negligible, indirect Minor
• Civil rights: direct Negligible, indirect Negligible
• Democracy: direct Negligible, indirect Negligible
• Privacy: direct Negligible, indirect Substantial
• Psychological: direct Negligible, indirect Negligible
• Epistemic: direct Minor, indirect Minor
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

Financial

Reported: Yes. The report explicitly describes individual victim losses reaching beyond US$40,000 in documented cases of tech support scams.

Indirectly caused: Victims of the tech support scams suffered documented financial losses exceeding US$40,000 per individual case after being routed through AI-enhanced lure pages.

Inferred additional harm: Given the 15,500 domains and numerous investment scams, we infer total financial losses across all victims likely reach tens of millions of dollars globally.

Malicious content

Reported: Yes. The report describes the creation and distribution of malicious scam content, deepfakes, and fake news articles.

Directly caused: Generative AI tools like DALL-E and deepfake video generators were used to directly create deceptive images, fake news anchor videos, and localized ad copy.

Indirectly caused: The malicious content was distributed at scale across advertising networks, compromised websites, spam emails, and social media platforms.

Inferred additional harm: We infer that millions of internet users were exposed to these AI-generated malicious ads and deceptive landing pages across global ad networks.

Differential treatment

Reported: Yes. The report describes targeting specific demographics, such as senior citizens and those in need of social assistance.

Indirectly caused: The WickedWally campaign specifically targeted vulnerable groups like senior citizens and individuals seeking social assistance with deceptive debt relief and Medicare lures.

Inferred additional harm: We infer that other vulnerable demographics, such as non-native English speakers or those with lower digital literacy, were disproportionately targeted and harmed by these localized campaigns.

Privacy

Reported: Yes. The report describes phishing sites and web forms designed to harvest personal information and credit card data.

Indirectly caused: The campaigns used web forms and fake chatbots to collect victims' contact details, personal information, and credit card data.

Inferred additional harm: We infer that the harvested personal data was likely sold on the dark web or used for subsequent identity theft and targeted phishing attacks.

Epistemic

Reported: Yes. The report describes the creation of fake news replica sites and deepfake news broadcasts to fabricate endorsements.

Directly caused: AI was used to generate deepfake videos of news anchors and public figures to fabricate endorsements for fraudulent platforms.

Indirectly caused: The dissemination of these deepfakes on replica news sites eroded trust in legitimate media organizations and public figures.

Inferred additional harm: We infer that the widespread use of high-quality deepfakes of real public figures and news outlets contributes to a broader erosion of public trust in digital information and journalism.

People affected

• Occurrences reported: 1
• People reportedly harmed: 2
• People reportedly exposed: 1000000

Potential causes

Management

• Inadequate Ad Platform Review: Ad networks fail to detect AI scams due to weak automated review tools.
• Abuse of Commercial Trackers: Keitaro tracker licensed to threat actors or run via illicit copies.
• Slow Abuse Mitigation: Security tools struggle to keep up with the scale of AI domain rotation.

Technology

• Generative AI Lure Generation: Generative AI used to mass-produce convincing scam headlines and copy.
• Deepfake Video Generation: AI-generated deepfakes impersonate anchors to build false trust.
• AI Text-to-Image Generators: DALL-E used to create realistic assets for tech support scam lures.

Data Inputs

• Deepfake Source Material: Using real video and audio of public figures to generate deepfakes.
• AI-Themed Marketing Lures: Promoting fake AI trading algorithms as data inputs to lure victims.
• Impersonated Brand Assets: Using popular brand logos and names as inputs for AI-generated ads.

Human Factors

• Overtrust in AI Technology: Victims trust fake claims of highly profitable AI-driven trading.
• Susceptibility to Deepfakes: Users find highly realistic deepfake video ads convincing and urgent.
• Susceptibility to Chatbots: Victims trust low-friction fake AI chatbots that qualify them for scams.

Process and Methods

• Evasion via Domain Cloaking: Using Keitaro TDS to hide AI scam content from security reviewers.
• Scalable AI Lure Production: Programmatic AI asset generation allows rapid scaling of scam pages.
• Targeted Ad Distribution: Using Bigo and Facebook Ads to deliver AI deepfakes to target audiences.

Regulatory Environment

• Lack of Deepfake Regulation: Absence of strict controls on creating and distributing deepfakes.
• Weak Ad Network Verification: Ad platforms fail to verify the authenticity of AI-generated ads.

Information quality

• Classification confidence: High
• Reason for confidence: The reports provide detailed, empirical cybersecurity research from Infoblox and Confiant, tracking over 15,500 domains and documenting specific threat actor groups and their TTPs. The role of AI in generating deepfakes, copy, and images is clearly documented, and the connection to the Keitaro TDS is thoroughly analyzed.
• Ambiguities identified: The exact generative AI models used for the deepfake video and text generation (other than DALL-E) are not specified.
• Alternative interpretations: The incident could be viewed primarily as a traditional cybercrime/phishing campaign where AI is merely an incidental tool, rather than an inherent failure of an AI system itself.

Threat actors leveraged generative AI and deepfakes of public figures, including the Canadian Prime Minister, to scale financial fraud across 15,500 domains. While the campaign demonstrates a highly sophisticated and evolved cybercrime capability, its impact is classified as minor from a national security perspective as it does not target critical infrastructure or systemic state functions.

• Overall national security impact: Minor
• Scope: Multiple nations
• Primary target: No clear primary
• Other affected: Canada, United States, Germany, Japan, Italy, Netherlands, France, Spain, Portugal
• Alleged perpetrator: FaiKast and WickedWally

Threat characteristics

• Imminence: Long-term. Represents an ongoing strategic cybercrime capability requiring continuous defensive monitoring rather than an immediate national security crisis.
• Autonomy: Human-controlled. Generative AI systems (DALL-E and deepfake generators) were utilized as static tools by human threat actors to create deceptive assets.
• Novelty: Evolved capability. Represents a significant advancement in existing phishing and fraud methodologies by integrating generative AI assets with sophisticated traffic distribution systems.

Impact by dimension

• Physical security: Negligible. No physical systems, critical infrastructure, or human safety elements were targeted or affected by this campaign.
• Information security: Minor. Involves automated generation of deepfakes impersonating public figures like the Canadian Prime Minister, representing an evolved but minor-scale threat to information integrity.
• Sovereignty: Minor. Impersonation of a G7 head of state (Canadian PM) to facilitate fraud represents a minor impact on government reputation, but did not disrupt core constitutional processes.
• Economic security: Minor. While the campaign targeted global citizens and caused individual losses over USD 40,000, it represents localized cybercrime rather than a systemic threat to national financial systems.
• Societal stability: Minor. Targeted vulnerable demographics like senior citizens with deceptive Medicare and debt relief scams, causing localized distress but no large-scale civil unrest.