/
Navigation
Chronicles
Browse all articles
Explore
Semantic exploration
Research
Entity momentum
Nexus
Correlations & relationships
Story Arc
Topic evolution
Drift Map
Semantic trajectory animation
Posts
Analysis & commentary
Pulse API
Tech news intelligence API
Browse
Entities
Companies, people, products, technologies
Domains
Browse by publication source
Handles
Browse by social media handle
Detection
Concept Search
Semantic similarity search
High Impact Stories
Top coverage by position
Sentiment Analysis
Positive/negative coverage
Anomaly Detection
Unusual coverage patterns
Analysis
Rivalry Report
Compare two entities head-to-head
Semantic Pivots
Narrative discontinuities
Crisis Response
Event recovery patterns
Connected
Search: /
Command: ⌘K
Embeddings: large
TEXXR

Chronicles

The story behind the story

days · browse · Enter similar · o open

Palo Alto Networks says in its testing, three weeks of frontier AI-assisted analysis matched a full year of manual penetration testing, with broader coverage

For the last several months, we have had early, unbounded access to the latest frontier AI models.

Palo Alto Networks Blog Sam Rubin

Context & Ripple Effects

Palo Alto Networks had already launched Cortex AgentiX, positioning AI agents as tools that can operate across security vendors' platforms. Its new testing claim extends that positioning from automated response into vulnerability discovery and penetration-testing work.

The result also arrives alongside Stanford's Artemis report, which found an AI hacking bot outperforming most tested human penetration testers in a university network. Together, the coverage points to AI moving from a security-assistance feature toward a more capable operator in security workflows.

First-order effects

  • Palo Alto Networks gains an internal performance claim for using frontier models to accelerate penetration-testing analysis while widening the systems and issues reviewed; the result remains a company test rather than an independently comparable benchmark.
  • Security teams using AI-assisted testing can assess whether to shift analysts from exhaustive manual discovery toward validating, prioritizing, and remediating AI-generated findings.

Second-order effects

  • Security-platform competitors will face pressure to pair AI agents for response automation with credible offensive-security and assessment capabilities, rather than treating AI as a standalone copilot feature.
  • Demand may shift toward tools that connect AI-agent activity with controls, governance, and cross-vendor workflows—an area Palo Alto had already targeted with Cortex AgentiX.

Third-order effects

  • If similar results hold across varied environments, penetration testing could become a continuous, AI-assisted process instead of a primarily periodic, labor-intensive engagement, changing the mix of work performed by security specialists.
  • The same capability raises the importance of validation and controls: broader automated discovery can improve coverage, but organizations will need to distinguish useful findings from model-generated noise and manage how far agents may act.

The trend: This is one data point in the shift from AI as a security analyst aid to AI agents performing larger portions of security discovery, triage, and response workflows.

Discussion

  • @peterwildeford Peter Wildeford on x
    PALO ALTO NETWORKS on MYTHOS: “In our testing, three weeks of model-assisted analysis matched a full year of manual penetration testing, with broader coverage.” [image]
  • @emollick Ethan Mollick on x
    So Mythos was, indeed, not marketing hype. Remember this is a general purpose model that just happens to be good at finding exploits because good models are good at lots of things. Expect similar from OpenAI & Google. And from open models in 8 months. https://hacks.mozilla.org/..…
  • Karthik Subramanian Karthik Subramanian on linkedin
    The latest frontier models, including OpenAI's GPT-5.5-Cyber, Anthropic's Mythos and Claude Opus 4.7, and the specialized variants emerging across major labs …
  • Sam Rubin Sam Rubin on linkedin
    As frontier AI models compress attack lifecycles from days to minutes, defenders must move even faster.  —  But security requires more than a model. …