Perplexity open sources numbat to monitor risky ai coding agents

Perplexity has released Numbat, an open-source tool designed to monitor the activities of AI coding agents on local endpoints. The utility provides detection and opt-in blocking capabilities to prevent unauthorised or dangerous actions by autonomous systems. This follows growing security concerns regarding the direct access AI agents have to development environments.

For enterprise teams deploying autonomous agents, Numbat offers a layer of observability and safety to mitigate risks like data exfiltration. It addresses the critical trust gap between agentic capabilities and strict enterprise security requirements.

  • Numbat monitors agent activity on endpoints to detect potentially malicious behaviour
  • It provides developers with the ability to opt in to blocking specific risky actions
  • The tool is open source to allow for community contributions and transparent security auditing
AI Agents & Automation Generative AI Custom Software
All AI news

More AI news

Models

OpenAI reports its Astra model can autonomously exploit unknown software vulnerabilities

OpenAI has disclosed that its Astra model is the first to reach a critical cybersecurity threshold. The system demonstrated the ability to identify and exploit previously unknown software flaws without any human intervention. This development represents a significant advancement in the autonomous capabilities of large language models within complex security environments.

Models

OpenAI releases GPT-6 Astra as its most powerful model to date

OpenAI has launched GPT-6 Astra, a new flagship model that the company describes as its most capable release. Chief executive Sam Altman has positioned the model as a leading benchmark for global AI performance across various sectors.

Models

Muse spark 1.3 (max) enters the top ten on the sevenlab ai leaderboard

Meta's latest model, Muse Spark 1.3 (max), has officially entered the top ten on the SevenLab AI leaderboard. Currently ranked at number seven, the model demonstrates significant performance gains within our value-adjusted rankings. This data is derived from ArtificialAnalysis metrics to help teams identify high-performing models.

Ready to build something
extraordinary?

15 minutes. No pitch deck. Just a conversation about what AI can do for your team.

Talk directly with our AI specialists

15 min, no strings
No sales pressure
Prototype in 7 days