# Organizations Apply AI to Cybersecurity

The discussion identifies current cybersecurity uses for AI—including security research, source-code analysis, penetration testing, and potential phishing defense—while highlighting limits in autonomous defensive remediation. AI can also produce insecure code or amplify user mistakes. Near-term operational activity is focused on controlling how employees use AI applications through approved-app policies, blocking or allowing other services, and DLP/CASB controls; the signal identifies no government agency, solicitation, contract, or procurement contact. For procurement professionals, this is a technology-evaluation and security-governance consideration rather than a specific buying opportunity.

- Acquisition teams evaluating AI-enabled security products can assess performance against defined tasks such as code analysis and testing, while accounting for the reported limits on autonomous remediation and the need for human oversight.
- Organizations deploying AI applications can use sanctioning controls and DLP/CASB to manage access and data uploads, as described in the signal.
- The discussion also points to faster AI-enabled attacks as a planning concern; security teams and prospective suppliers should consider whether proposed capabilities can support defensive teams without relying on autonomous action.

**Jurisdictions:** federal
**Industries:** Information Technology
**Topics:** Cybersecurity, Artificial Intelligence
**Published:** October 10, 2026

### Key Quotes
> With blue team, its far from ready due to a million midas touch like problems that need fixing. Think of asking "make this computer safe from the internet" and the agent just disables the NIC.
> — Community commenter (username not provided)

> Today most of what we do is practical, figure out which AI apps are sanctioned, block or allow the rest, and use DLP/CASB to see how people are actually using them, like stopping uploads to the free version of ChatGPT.
> — Community commenter (MSP employee; username not provided)

### Sources
- [What can AI currently do and not do in cybersecurity?](https://www.reddit.com/r/cybersecurity/comments/1x2bxyp/what_can_ai_currently_do_and_not_do_in) - reddit-cybersecurity