Topic · Prompt injection
Prompt injection attacks: incidents, defenses and definitions
cisonews.ai has logged 7 publicly reported prompt-injection incidents, with event dates from the week of September 12, 2022 to October 26, 2026.
The most recent is "In-the-wild attack plants malicious AGENTS.md file to hijack AI coding agents" (affected: AI coding agents (OpenAI Codex, Anthropic Claude Code)); each entry below links to its sourced write-up, followed by 3 explainers.
- 7
- Incidents
- 3
- Explainers
- 3
- Definitions
Collected from 13 pages on this site, newest addition October 27, 2026. Every fact sits on the linked page with its sources.
Incidents, newest first
Full log →- 2026-10-26In-the-wild attack plants malicious AGENTS.md file to hijack AI coding agentsAffected: AI coding agents (OpenAI Codex, Anthropic Claude Code)
- 2026-06-04SearchLeak flaw, CVE-2026-42824, disclosed in Microsoft 365 CopilotAffected: Microsoft 365 Copilot
- 2025-08-12GitHub Copilot and Visual Studio prompt injection code execution flaw (CVE-2025-53773)Affected: GitHub Copilot and Visual Studio
- 2025-07-08Cursor AI code editor patches CurXecute prompt-injection RCE flawAffected: Cursor AI code editor (Anysphere)
- 2025-06-11EchoLeak: zero-click prompt injection flaw in Microsoft 365 Copilot (CVE-2025-32711)Affected: Microsoft 365 Copilot
- 2024-08-20Slack AI indirect prompt injection that could pull data from private channelsAffected: Slack AI
- 2022-09-12Twitter users hijack Remoteli.io's GPT-3 job bot with prompt injectionAffected: Remoteli.io Twitter bot (OpenAI GPT-3)
Questions about this topic
- How many prompt-injection incidents has cisonews.ai logged?
- 7, with event dates from the week of September 12, 2022 to October 26, 2026. The most recent is "In-the-wild attack plants malicious AGENTS.md file to hijack AI coding agents".
- Which MITRE ATLAS techniques appear in these incidents?
- By this site's own tagging, the most frequent are LLM Data Leakage (AML.T0057, 2 incidents); LLM Prompt Injection (AML.T0051, 2 incidents); LLM Prompt Injection: Indirect (AML.T0051.001, 2 incidents).