Every AI model should be treated as an insider risk, says security founder on data poisoning
AccBalanced · x · 2026-10-27
A discussion highlighting that every AI model — including those from US labs — is susceptible to data poisoning, since models train on data they didn't create: scraped from random internet sources, provided by clients, or purchased from scrapers and data buyers. The author (KindoAI founder) argues every model should always be treated like an insider risk and properly secured, which is why he built KindoAI.
More from Safety
- Agent "hives" shift the attack surface from models to permissions and keys — AuriumD · 2026-10-27
- Dev flags 'concerning' Cohere open-model behavior: system-wide scan and data upload to US entity — basedjensen · 2026-10-27
- Atomadic ships starter MCP with 50 tools for AI compliance, identity, escrow and EU AI Act — modelcontextprotocol · 2026-10-27
- Off-grid data center gas power emits 63-80% more than the grid — 2C_ornot2C · 2026-10-27
- AI is making hard-to-exploit bugs practical: PHP2Shell as proof — AccBalanced · 2026-10-27
- LiteLLM supply-chain attack: malware steals credentials by just starting Python — AccBalanced · 2026-10-27