12.03
Agent security
You treat everything the agent reads as untrusted data rather than a command, and reason about the lethal trifecta, prompt injection, tool misuse, permission escalation and memory poisoning as an agent's own attack surface, distinct from generic responsible-AI.
- Defining AI Guardrails and Agent Security
- 18 min
- Pro
Who's hiring for this skill
79 companies have live AI-PM roles that ask for skills taught in this lesson.
- Anthropic
- Sierra
- Glean
- Decagon
- Scale AI
- Datadog
- Nebius
- Harvey
- OpenAI
- GitLab
- Snowflake
- Stripe
- Okta
- Databricks
- Intercom
- Lyft
- Vanta
- Asana
- Fivetran
- Gusto
- Writer
- Cohere
- Brex
- Coinbase
- CoreWeave
- Coursera
- Cresta
- Fireworks
- Grafana
- JetBrains
- Mercor
- Mistral
- Rogo
- Temporal
- Vercel
- 1Password
- Box
- Calendly
- Deepgram
- Docker
- Lightning Ai
- Luma AI
- Notion
- Plaid
- Snorkel AI
- Thinking Machines
- Typeface
- Lovable
- Perplexity
- Ramp
- Replit
- Baseten
- Bland Ai
- Cerebras
- Customer.io
- Discord
- Drata
- Dust
- Eleven
- ElevenLabs
- Eve
- Klaviyo
- LangChain
- LaunchDarkly
- Legora
- Make
- Nabla
- Render
- Sardine
- Sigma
- Sourcegraph
- Speak
- Spotify
- Synthesia
- Together AI
- Vapi
- Weaviate
This subchapter is part of the paid course.
Its video, its written lesson, its exercise and its grading are all included on any plan. The free subchapters stay free either way.