AIToday
Large Language ModelsAI Safety & AlignmentImpress WatchPublished: Oct 22, 2026, 06:01 JST

GPT-5.6 Sol found the critical PHP flaw, OpenAI says

Send on LINE
GPT-5.6 Sol found the critical PHP flaw, OpenAI says

3 Key Points

  1. What happened

    OpenAI said its GPT-5.6 Sol identified the critical PHP flaw as part of its Patch the Planet initiative; security firm Calif used it for the find, and Google-owned Wiz independently found and reported both vulnerabilities.

  2. Why it matters

    The flaw's severity and PHP's wide use make it "one of the most impactful security issues in recent years" by OpenAI's account, and OpenAI warns the same AI capability could speed attackers' exploit development.

WHO IT HITSOperations and security teams running PHP-based services face priority patching, and OpenAI is offering a PHP version checker plus a planned Codex skill to help them identify exposed versions and decide which updates to apply.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, in your inbox every morning.

Context & Analysis

The disclosure lands alongside a practical disruption: many internet services at home and abroad stopped around the 11pm security update on the 21st, as operators moved to apply fixes for the PHP flaw. OpenAI frames the find as one of several critical vulnerabilities it has already identified and is working to fix, and it expects similar discoveries to surge in the coming months — a pace that explains its push for a shared approach.

That approach is being assembled from several directions at once. PHP's maintainers led the fix and release, while OpenAI's team worked with Wiz, Cloudflare and other partners on investigation, technical support and coordination to prepare protections ahead of publication. Calif's use of GPT-5.6 Sol came under Patch the Planet, one of the initiatives OpenAI cites.

OpenAI's stated worry is symmetric: the same capability that helps defenders can also speed attackers' development of attack tools and exploits. Alongside the version checker and planned Codex skill, it points to support such as Daybreak for Frontline Defenders as a way to put frontier models in more defenders' hands, and it is asking the security community to improve and cooperate on the framework it plans to share.

FAQ
Who found the PHP vulnerability?
OpenAI says its GPT-5.6 Sol identified it, with security firm Calif using the model under the Patch the Planet initiative. Google-owned Wiz independently found and reported both vulnerabilities.
What is OpenAI offering to help companies patch?
OpenAI is providing a PHP version checker and plans to release a Codex skill, so operations staff can spot potentially vulnerable PHP versions and judge which updates to apply.
Does OpenAI expect more findings like this?
Yes. OpenAI calls this one of several critical vulnerabilities it has already found and is working to fix, and it expects similar discoveries to surge in the coming months.
Send on LINE

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.

Questions and answers are published on this page.

Related Articles

Next articleMeta revenue hits $64.6 billion but $10 billion settlement halves margin