In Other News: PromptPwnd Attack, Small macOS Bounties, Chinese Hackers Trained in Cisco Academy

Aikido Security has uncovered a new prompt injection attack method dubbed PromptPwnd, which leverages GitHub Actions and AI agents by embedding malicious prompts into issue bodies, commit messages and PR descriptions that AI systems such as Gemini CLI, Claude Code, OpenAI Codex and GitHub AI Inference interpret as instructions, with at least five Fortune 500 companies reported as affected and Google patching Gemini CLI within days of notification.

The round-up also notes that two individuals from China, who were highly successful Cisco Network Academy Cup students in 2012, later became key operators of the APT group Salt Typhoon, with SentinelOne noting their early education on Cisco products likely aided their expansive intelligence‑gathering operations across more than 80 telecommunications companies globally.

The piece highlights the breadth of security research and industry responses surrounding these developments, including prompt injection risks and post‑quantum cryptography initiatives mentioned elsewhere in SecurityWeek’s coverage. It also briefly touches on ongoing discussions about macOS bug bounties and related security incentives. In conclusion, SecurityWeek’s “In Other News” roundup underscores how education, attacker tools and disclosure timelines intersect to shape threat landscapes and defensive postures.