NEW DELHI — Researchers at OpenAI and partner organizations have used advanced artificial intelligence models to uncover critical security flaws in PHP, the foundational programming language that powers nearly 70 percent of websites worldwide. The discovery has prompted an urgent push across the technology industry to validate and deploy software fixes before malicious actors can exploit the weaknesses. With artificial intelligence demonstrating an unprecedented ability to analyze complex code, defenders are now in a high-stakes race to safeguard core internet infrastructure against an emerging wave of automated cyber threats.
Critical PHP Flaws Detected by Project Daybreak
The vulnerabilities were uncovered by researchers associated with Project Daybreak, a dedicated cybersecurity initiative at OpenAI, utilizing the company’s latest artificial intelligence model. Because PHP underpins a vast majority of online services, flaws within its framework pose widespread risks to digital systems globally.
Led by Clint Gibler of OpenAI’s cybersecurity team, specialists spent more than two months verifying the findings and evaluating potential remedies to ensure the flaws were addressed prior to public disclosure. The urgency of the operation was underscored by reports that millions of exploitation attempts occurred during this validation period, demonstrating the speed at which attackers seek to capitalize on underlying software weaknesses.
AI Transforms Vulnerability Detection and Cyber Offense
The PHP discovery marks a key milestone in the application of artificial intelligence to cybersecurity, joining broader industry efforts where models from OpenAI, Anthropic, and newer firms such as Depthfirst have identified weaknesses in web browsers and internet software. While these capabilities allow defensive teams to detect flaws and draft patches far faster than manual methods allow, they also present severe risks if turned toward offensive operations.
Security professionals are particularly concerned about the proliferation of powerful, accessible tools, including Chinese open-weight systems like DeepSeek and Kimi. Experts warn that the availability of such technologies could significantly lower the barrier for attackers looking to discover flaws and execute breaches at scale.
Malicious Actors Deploy AI Agents in Global Attacks
Recent security findings reveal that artificial intelligence is already assisting active cyber operations. Reports indicate that threat actors have employed models such as Anthropic’s Claude and DeepSeek to illicitly capture hundreds of thousands of payment card records and compromise roughly 100 websites.
Furthermore, Microsoft recently disclosed that Chinese hackers utilized an AI agent tool known as Artex, originally designed for security testing, to target hundreds of entities, including government and military institutions. As offensive automation shifts from theoretical risk to operational reality, software developers and infrastructure providers must focus on rapidly deploying, verifying, and maintaining security updates to ensure artificial intelligence strengthens digital defenses rather than undermining them.
Follow for daily updates on cybercrime, corporate fraud, DFIR, hacking, investigations, and digital forensics
Leave a Reply