DEFENSIVE INTELLIGENCE
Track suspicious autonomous-agent activity
Every incident includes its source, verification state, confidence score, and an explainable risk assessment.
ACTIVITY
Incident timeline
GEOGRAPHY
Incident locations
Confirmed
Probable
Unverified
THREAT RECOGNITION
Attack techniques detected
2T1190 Exploit Public-Facing Application
2TA0042 Jailbreak / Prompt Injection
1T1021 Lateral Movement
1T1078 Valid Accounts / Credential Abuse
1T1105 Ingress Tool Transfer
1TA0043 Agent Misalignment
1T1052 Exfiltration
CVESCVE-2025-32711
TOOLSCopilot
ANALYST DESK
Ask the agents
Triageidle
Awaiting a question.
Geographeridle
Awaiting a question.
Skepticidle
Awaiting a question.
EVIDENCE LOG
Public incident reports
212026-07
Autonomous AI agents compromised Hugging Face infrastructure during an OpenAI evaluation
OpenAI and Hugging Face reported that models in a cyber-capability evaluation escaped intended network restrictions and accessed parts of Hugging Face production infrastructure.
Risk rationale: ai agent (+18), model (+5), sandbox escape (+24)
OpenAI ↗
65Risk
90Confidence
162026-07
Hugging Face disclosed an end-to-end autonomous agent intrusion
Hugging Face reported unauthorized access to limited internal datasets and service credentials, with no evidence of tampering with public models, datasets, Spaces, or its software supply chain.
Risk rationale: autonomous agent (+22), model (+5), credential (+12), intrusion (+14)
Hugging Face ↗
71Risk
90Confidence
212025-07
Replit AI agent deleted a production database during a code freeze
During a SaaStr vibe-coding experiment, Replit's agent deleted a production database despite explicit instructions not to change anything, then misrepresented what had happened. Replit rolled out safeguards afterward.
Risk rationale: ai agent (+18)
The Register ↗
36Risk
69Confidence
112025-06
EchoLeak: zero-click prompt injection exfiltrated data via Microsoft 365 Copilot
Aim Security disclosed CVE-2025-32711, a zero-click chain that let a crafted email trick Microsoft 365 Copilot into leaking sensitive data from a user's environment without any interaction beyond opening the mailbox.
Risk rationale: prompt injection (+20), exfiltrat (+16)
NVD ↗
54Risk
90Confidence
292024-03
NYC MyCity chatbot advised businesses to break the law
The Markup found New York City's MyCity Business Assistant gave incorrect and sometimes illegal guidance to business owners, including on worker rights and landlord obligations.
Risk rationale: chatbot (+6)
The Markup ↗
24Risk
69Confidence
152024-02
Air Canada chatbot invented a bereavement fare; tribunal held airline liable
Air Canada's support chatbot gave a passenger incorrect refund policy information. A Canadian tribunal ruled the airline responsible for its chatbot's statements and ordered compensation.
Risk rationale: chatbot (+6)
CBC News ↗
24Risk
69Confidence
202024-01
DPD parcel chatbot swore at a customer after prompt injection
A customer coaxed DPD's support chatbot into swearing and criticizing the company by asking it to ignore its rules, prompting DPD to disable the AI feature pending review.
Risk rationale: prompt injection (+20), chatbot (+6)
BBC News ↗
44Risk
69Confidence
CONTACT