Google Gemini vulnerable to ASCII smuggling attacks, no fix planned
A cybersecurity researcher discovered that Google Gemini is vulnerable to ASCII smuggling attacks, where hidden characters in text can be used as commands to manipulate the AI's behavior. Google stated it does not consider this a security bug and has no plans to fix it, dismissing the vulnerability as a social engineering issue rather than a technical flaw.
What happened
Cybersecurity researcher Viktor Markopoulos from FireTail discovered an 'ASCII smuggling' attack vulnerability in Google Gemini that involves hidden characters in text which the AI reads as commands but humans cannot see.
Markopoulos demonstrated that invisible control characters or Unicode symbols could be weaponized through text-based inputs like calendar invites or emails to make Gemini misinterpret or rewrite data, change meeting details, or generate misleading outputs.
When the same attack was tested against OpenAI's ChatGPT, Anthropic's Claude, and Microsoft's Copilot, those models either sanitized or rejected the hidden inputs, but Gemini, along with Elon Musk's Grok and China's DeepSeek, failed to block them.
Reported impact
Because Gemini works closely with email, scheduling, and document systems, there is a risk that it might expose confidential information or help spread misinformation in corporate networks.
Index analysis
Google classified ASCII smuggling as a 'social engineering' problem rather than a technical vulnerability, and stated it does not consider this a true security bug.
Governance controls
Google has said it has no plans to issue a fix for the flaw.
Classification
Sources and evidence
Sources not yet published
This record was researched and written by the Index from original reporting. Its sources are still being verified by an editor, and they are listed here once that check is complete. Until then the record carries no reviewer.