OpenAI Fixed Critical PHP Vulnerabilities

OpenAI announced the completion of a two-month coordinated program to address critical vulnerabilities in the PHP programming language. A significant portion of websites on the internet run on this language, making the issue large-scale.

The first critical vulnerability was discovered by Calif using the GPT-5.6 Sol model. Independently, researcher Wiz identified additional vulnerabilities.

Patches were released by the PHP maintainers themselves. Major providers, including Cloudflare, learned about the issue in advance and managed to prepare protection for their users.

OpenAI calls this one of the most serious security issues in recent times. The official announcement does not disclose technical details of the bugs themselves.

Website owners are recommended to take the following actions:

• Update the PHP version. You can check your current version using the OpenAI php2shell-checker utility on GitHub.
• Consult the instructions from the language maintainers and your hosting provider.
• Review the summary information on the Patch the Planet initiative website.

US government agencies, including CISA and the White House, participated in the coordinated closure of vulnerabilities.

OpenAI acknowledges that the same AI capabilities help attackers quickly write exploits. The company forecasts a sharp increase in such findings in the coming months.

To protect the industry, the company is allocating $1 billion in subsidized access to its artificial intelligence models.