As of October 22, 2026, Level lists 197 live jobs whose title names AI security, from 75 employers. 86% of them sit at Works on AI or Builds AI. This week, AI models helped find critical flaws in PHP, and a compromised AI library shipped malware. Here is who is hiring to defend against both.
What happened
On October 21, 2026, the PHP project shipped versions 8.5.12, 8.4.27, 8.3.36 and 8.2.35. The PHP Foundation says they fix two critical vulnerabilities. Under the right conditions, both let an unauthenticated attacker run code on a server.
Wiz Research named the issue PHP2Shell. Its post says nearly every PHP version is affected by at least one flaw, back to PHP 5.4. Wiz writes that frontier AI models made the exploit chain practical to demonstrate.
OpenAI published its side the same day, under the name Patch the Planet. It says a firm called Calif found a critical flaw using one of its models. It also commits $1 billion in subsidized Daybreak access for organizations with limited security resources. OpenAI expects more findings of this kind.
A second story hit the same day. SafeDep reported that the LiteLLM 1.107.0 release on PyPI carries a credential stealer and a backdoor. LiteLLM is a gateway that lets code call 100+ LLM APIs through one format. SafeDep says the upstream GitHub source does not contain the malicious files.
Both stories drew hundreds of points on Hacker News, for PHP2Shell and for LiteLLM. Together, they show AI finding flaws and AI tooling carrying malware.
Who is hiring for AI security jobs right now?
Level counts 197 live roles with AI, ML, LLM or agentic security in the job title. They come from 75 employers.
Amazon has the most, with 27 open roles. NVIDIA follows with 12. Cisco, Abnormal Security and Zscaler each list 9. State Street, Gartner, Thales and Cato Networks are also in the top ten.
The mix is broader than security vendors. Chip makers, banks and research firms are all hiring for the same skill. You can browse the wider set on the security jobs page.
Are these real AI jobs or security jobs with an AI label?
Level grades every posting on how much AI work it involves. The four levels are Builds AI, Works on AI, Uses AI and Little AI. The levels page explains how each one is assigned.
Of these AI security roles, 169 sit at Works on AI or Builds AI. That is far above the catalog as a whole. Level has 169,000+ live jobs in total. Only 12% of them reach those two levels.
The remaining 28 sit at Uses AI or Little AI. The top level alone covers 69 roles. That is 35% of the set. You can see that tier on the Builds AI page.
The broader security category tells a different story. Level has 2,800+ live security jobs in total. Only 11% of them reach Works on AI or above. So most security work is still classic security, and the AI share is a small, specialized slice.
What do AI security jobs pay, and how senior are they?
Just over half of these roles publish a salary in US dollars, 102 in total. The median posted minimum is $168,000 a year. The median posted maximum is $238,250.
These are senior roles. 85% of them are tagged senior, lead or executive. Only 5 are tagged intern or junior.
Engineering titles dominate. 55% of the titles include the word engineer. Twelve titles mention research. Fifteen mention architect.
Remote work is possible but not the norm. 38% of these roles are listed as remote. The remote AI jobs page shows the full remote set.
For a candidate, the takeaway is practical. Most openings sit at senior level or above. If you are earlier in your career, the cybersecurity outlook covers where entry points still exist.
Open roles right now
Level lists 309 live security jobs at Works on AI or above. Here are ten of them.
