Skip to content
Submit incident
AII-1000-70081
Under investigationA regulator, court or police has opened a formal inquiry and reached no finding yet.

Anthropic's Mythos AI raises cybersecurity concerns for SEC database

The American Securities Association warned that Anthropic's Mythos AI model, designed to find code vulnerabilities, could be exploited to breach the SEC's Consolidated Audit Trail database and expose retail investors' personal and trading data. U.S. Treasury and Federal Reserve officials convened bank CEOs to discuss the risks.

16 April 2026 Β· Anthropic Β· Security and cyberattacks
16 April 2026
AII-1000-700812 sources ↓
Quick facts
Category
Security and cyberattacks
Country
United States
Source
Business Insider
Date
16 April 2026
Drafted by
Raidu Incident Detection

What happened

Anthropic announced Claude Mythos in early April 2026 as part of Project Glasswing, a cybersecurity initiative. The model is designed to locate thousands of high-severity vulnerabilities in code across major operating systems and web browsers.

The American Securities Association publicly warned that Mythos could be exploited by malicious actors to breach the SEC's Consolidated Audit Trail (CAT), a database containing retail investors' private information including personal details, trading strategies, and portfolio positions.

Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell convened major U.S. bank CEOs for an urgent meeting about the cyber risks posed by Mythos.

Reported impact

The ASA outlined that malicious use of Mythos could lead to mass identity theft of retail investors in a single breach and exposure of sensitive trading strategy and portfolio data.

The group warned that the combination of large-scale financial data and increasingly capable AI systems could create new vulnerabilities that regulators may not be fully prepared for.

Index analysis

The CAT was designed to give regulators a detailed view of trading activity across U.S. markets for surveillance and investigations, but its design does not account for AI-driven threats now emerging.

Even anonymised datasets processed with modern AI tools can carry strategic value when extracted by malicious actors, according to expert analysis.

Governance controls

The ASA called for immediate safeguards, including halting collection of personal data from retail investors and deletion of existing records stored in the CAT.

Classification

Category
Security and cyberattacks
Industry
Financial and insurance services
Country
United States
Harm status
Potential harm
Intent
Unintentional
Timing
Unknown
Harm types
Economic or property, Reputational, Public interest
Harmed
Consumers, General public
AI systems
Mythos

Sources and evidence

This record was researched and written by the Index, based on the following original reporting.

Business Insider
Anthropic's Mythos AI tool could spark 'systemic financial market disruption,' trade group warns
16 April 2026
India Today
Anthropic's Mythos AI model could expose trading data, warns US financial group
17 April 2026