Research
FARMA attack poisons agent reasoning logs with forged decision traces
Researchers at Penn State have demonstrated a new memory-poisoning attack that forges an agent's own decision logs rather than its retrieved knowledge, achieving 100% success across multiple models.
1 min read
Sourcer/ai-agents
Ask AI about this articleResearchers at Penn State have published a novel attack on agentic systems that targets not what agents retrieve, but what they write about their own reasoning. The attack, called FARMA, exploits the decision logs and rationales that agents store to track their past work. Unlike prior memory-poisoni...
Sign in to read the full analysis
Free account. Full analysis on LLM unit economics, plus the weekly Cost-of-Inference column.
Try it on your own context
You just read the writeup. Now run the thing. Paste a doc or some verbose tool output and watch it shrink — free, no signup.
2,912/12,000 chars
Compressed
Compressed text will appear here…
Method & sources
- Source type
- Primary publication (lab/vendor blog) — our analysis + implication
- Source link
- r/ai-agents
- Published
- UTC
- Byline
- By the gotcontext.ai team (editorial standards)
- Correction?
- corrections@gotcontext.ai