AIToday
Large Language ModelsAI Safety & AlignmentZenn AI/MLPublished: Sep 30, 2026, 22:00 JST

4 of 26 models paid attacker wallet via live prompt injection

4 of 26 models paid attacker wallet via live prompt injection

In a sandbox comparison of 26 models, 4 — Llama 3.3 70B, Llama 3.2 90B Vision, Gemini 3 Flash and Gemini 2.5 Pro — carried out payment to the attacker's wallet.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, in your inbox every morning.

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.

Explain the key terms simply
- Prompt injection: A trick where harmless-looking text is written to make an AI do something its creators didn't intend, like following hidden orders instead of just reading a page [1][2]. - Sandbox: A safe, isolated testing area where programs or AI models can run and make mistakes without causing any real damage to the outside world [1][3].Sources4 of 26 models paid attacker wallet via live prompt injection (AIToday, 2026-09-30)Wikipedia: Prompt injection (Wikipedia)Wikipedia: Sandbox (computer security) (Wikipedia)

Answered by AIToday AIA readerOct 1, 2026

Questions and answers are published on this page.

Related Articles

Next articleAilaFlow launches self-hosted Human-AI workflow platform