Story perspectives

OpenAI and Anthropic Launch AI Tools to Combat Cyber Threats

4/16/2026

Summary

AI Firms Strengthen Cybersecurity

OpenAI has launched GPT-5.4-Cyber, a cybersecurity-focused variant, to strengthen defensive capabilities for thousands of verified users.
Anthropic has released Claude Mythos under Project Glasswing, which has identified numerous long-standing vulnerabilities in major software.
Both OpenAI and Anthropic aim to democratize access to advanced cybersecurity tools while implementing safeguards to prevent misuse.
Financial leaders discuss sector implications as concerns grow that these AI models could be exploited by malicious actors.

Summary

Anthropic's Vulnerability Initiative

Anthropic's Project Glasswing, powered by Claude Mythos, aims to identify software vulnerabilities across major systems.
The initiative involves over 50 partners, including tech giants Amazon and Google.
Mythos has reportedly found thousands of vulnerabilities, including long-standing flaws in OpenBSD and FFmpeg.
Critics question transparency, arguing the lack of public access may hinder validation.
Nevertheless, the project signals a shift in vulnerability discovery speed and raises patching concerns.

History

Historical Context

The Cybersecurity Act, enacted in 2015, aimed to enhance information sharing between the U.S. government and private sectors amid a surge in cyberattacks, which increased by 1,300% from 2006 to 2015. This law reflects ongoing efforts to protect critical infrastructure, paralleling advancements by companies like OpenAI and Anthropic, which address modern cybersecurity challenges while raising concerns about potential misuse of AI technologies.

1 of 3

AI Firms Strengthen Cybersecurity

OpenAI has launched GPT-5.4-Cyber, a cybersecurity-focused variant, to strengthen defensive capabilities for thousands of verified users.
Anthropic has released Claude Mythos under Project Glasswing, which has identified numerous long-standing vulnerabilities in major software.
Both OpenAI and Anthropic aim to democratize access to advanced cybersecurity tools while implementing safeguards to prevent misuse.
Financial leaders discuss sector implications as concerns grow that these AI models could be exploited by malicious actors.

1 / 3