Science & Tech · AI · 6 hrs ago
OpenAI launches initiative to find and fix software security flaws
Two serious flaws in PHP software could let hackers take over web servers.
PHP powers a large share of the world's websites, making the vulnerabilities known as PHP2Shell a risk to many systems.
The flaws were reportedly found using OpenAI's GPT-5.6 Sol, showing how AI can help identify subtle security problems.
OpenAI has launched Patch the Planet to coordinate efforts to find and fix vulnerabilities with cybersecurity professionals, affected organizations and government agencies.
The company says it will provide $1 billion in subsidized access to its Daybreak security-auditing platform for organizations with limited resources that protect essential services.
OpenAI also released a Codex tool on GitHub to check websites for signs of PHP2Shell compromise.
The company expects similar vulnerabilities to be found more often, and is inviting outside experts to review its approach and identify gaps.
OpenAI announced “Patch the Planet,” a framework intended to help cybersecurity professionals find and fix software vulnerabilities.
The initiative aims to coordinate work among security professionals, vulnerable organizations, government agencies and other concerned parties.
OpenAI said the effort responds to the risk that AI can help attackers develop tools and exploits as well as help defenders find flaws.
The company said it would devote $1 billion to subsidize access to its Daybreak cybersecurity auditing platform for organizations with limited security resources.
OpenAI also launched a Codex skill on GitHub to check websites for possible compromise by PHP2Shell.
- Who
- OpenAI launched the initiative and invited third-party cybersecurity professionals to help review it.
- What
- It introduced a framework called “Patch the Planet” to help find and fix software vulnerabilities.
- When
- Wednesday, October 21, 2026.
- Where
- The initiative is intended to foster global cooperation. No specific location is stated.
- Why
- OpenAI said AI can help uncover security flaws, but can also help attackers develop offensive tools and exploits.
This story does not have two clearly opposing sides.
OpenAI published a blog post saying its recent models had substantially improved at translating codebases between programming languages.
Reports emerged that emergency patches had been released for two major PHP vulnerabilities.
OpenAI announced the “Patch the Planet” framework after the initial reports of the patches.
- Initiative
- Patch the Planet
- Daybreak funding
- $1 billion for subsidized access for organizations with limited security resources
- PHP vulnerabilities
- Two major flaws, known as PHP2Shell
- Codex skill
- Launched on GitHub to check websites for PHP2Shell compromise
- Models cited
- GPT-6 Astra and GPT-5.6 Sol
What people are saying
to help turn findings into protection and build a response that can scale as AI capabilities advance
It highlights how AI can uncover security flaws with internet-scale impact, but the same capabilities can also help attackers rapidly develop offensive tools and exploits.
and we expect similar findings to increase sharply in the coming months