OpenAI GPT-4o guardrail jailbreak via zero-width Unicode characters
A guardrail jailbreak vulnerability affecting OpenAI's GPT-4o was discovered that uses zero-width Unicode character encoding to bypass restrictions on providing information hazards and illegal content. The vulnerability was assessed as posing minimal societal risk and limited primarily to nuisance-level policy breaches.
What happened
A guardrail jailbreak vulnerability was discovered in OpenAI's GPT-4o that uses zero-width Unicode character encoding and character manipulation to bypass inference restrictions on information hazards and legal violations.
Reported impact
The vulnerability was assessed as posing minimal societal risk, with impact limited primarily to nuisance or minor policy breaches affecting individual users without broader implications.
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.