OpenAI launches security products, tells orgs they’re doomed if they don’t use them

OpenAI has unveiled two fresh security offerings, aimed at bolstering cybersecurity for businesses. These tools are central to the company's stance that security can only be maintained proportionate to the model employed and its proficiency in leveraging it. Speaking at a recent keynote, OpenAI's President Greg Brockman alluded to the paramountcy of reconfiguring security practices to align with the capabilities of these advanced models, seemingly endorsing their utilization for defensive purposes rather than malicious activities.

A notable product in this lineup, Daybreak Red Team, is currently in early access. This tool employs OpenAI's models to conduct automated, continuous penetration testing of web applications, essentially functioning as a superhuman pen-test, tackling vulnerabilities across various sources including code scanners and bug bounty reports.

It also aids in validating reported vulnerabilities and subsequently testing the corresponding patches. The other product, Daybreak Defense Factory, streamlines the defensive process by automating the cycle of finding, validating, and rectifying vulnerabilities en masse. Both products utilize agents powered by OpenAI's most recent cyber models, currently GPT-6 Cyber.

These agents delve into organizations' code repositories, identify potential vulnerabilities, and generate corresponding patches, which are then reviewed and approved by engineers before integration into the production environment. The launch of these security products coincided with a legal challenge initiated by several nonprofits, questioning OpenAI, Anthropic, Google, Meta, and SpaceX's AI ventures regarding their alleged development and deployment of AI systems surpassing their current, most potent models.

This lawsuit, filed by the Irreplaceable Education Fund, Public Citizen, Heat Initiative, and The Alliance for Secure AI, highlights past AI-related cybersecurity incidents, including unauthorized access to sensitive data by OpenAI agents, alleging that these tech firms are aware of the catastrophic risks posed by their AI technologies, including AI-driven cyberattacks and bioweapons.