Microsoft warns of AI 'sleeper agents' hidden in model weights

Microsoft Research demonstrated a technique that could plant hidden backdoors in AI systems via model weights, but says it has not yet observed such 'sleeper agents' being used in the wild.

Read the full story at ITPro AI ↗

Timeline · 1 report

2026-11-12 11:30 · ITPro AI
Microsoft warns against AI ‘sleeper agents’ hidden in model weights

More stories

Microsoft Research shows AI agents can be backdoored by editing fewer than 10 neurons — Microsoft Research Blog
Llama App - Windows’ta ücretsiz indir ve yükle — Microsoft Research Blog
IBM and Microsoft Launch Imagine Enterprise AI Agent Platform — DEV Community — AI
Anthropic expands into enterprise productivity with Fable 5.2 and Mythos 5.2 amid IPO filing — DigiTimes Asia
Paul Weiss to open Silicon Valley office, hiring lawyer behind OpenAI's Microsoft and Alphabet deals — Reuters AI
AI-linked firms make up 71% of value of businesses led by Indian-origin executives — Mint AI
Microsoft's AI foundation models trace roots to OpenAI partnership and Azure infrastructure — InfoWorld AI
Why Google is chasing Jev into decision models — The Deep View

Get the daily brief of stories like this at 6:30 every morning →