GitHub Copilot: Sorry Dave, I can't do that harmful thing - unless you ask me in code

2026-07-08T19:23:58Z21484eb8f763acc05101ea4acf4fbd62f8623ba46c2ac2a5d72912501832bcbc
active-exploitationagentic-ransomwareai-jailbreakbackdoor-codebotnetcisa-kevclaudeconfidential-computingdata-leakdeepseekeviltokensgithub-ai-agentgithub-copilotllm-securitymfanetnutoracle-e-business-suitepegasusphishing-kitpredatorransomwaresharepoint-rceshinyhuntersspywarezero-day

What happened

A roundup of security stories highlighting widespread AI/LLM abuse and jailbreaks (GitHub Copilot, GitHub AI agent leaks, DeepSeek in‑browser ransomware, agentic ransomware, Claude backdoor concerns), multiple actively exploited software vulnerabilities (Oracle E‑Business Suite exploited pre-public exploit, SharePoint RCE added to CISA KEV), and significant data-breach and botnet incidents (Medtronic/ShinyHunters, Moody Bible Institute, NetNut takedown). Additional items cover phishing/tooling (EvilTokens), spyware/legal fallout (Predator/Pegasus), weakened MFA adoption, and broader concerns ~

Why it matters

A reviewed impact interpretation has not been published for this record.

Evidence and limitations

Source ID
theregister_security
Record identifier
21484eb8f763acc05101ea4acf4fbd62f8623ba46c2ac2a5d72912501832bcbc
Enrichment time
2026-07-08T19:23:58Z
AI-assisted enrichment
Yes

This record may overlap with other records. Its enrichment can be incomplete or wrong, and machine assistance was used. Validate consequential decisions against the linked source and your own environment.

Record · GitHub Copilot: Sorry Dave, I can't do that harmful thing - unless you ask me in code · Baitaphish