|
|
|
|
|
|
by rflsandroni
467 days ago
|
|
|
As AI agents and LLM-based systems are integrated into enterprise, finance, and customer-facing platforms, adversarial prompt threats have rapidly evolved. Today, ex-Apple Siri engineers are releasing ModernGuard, a 300M parameter multilingual transformer encoder, trained on millions of prompt attack simulations, built to detect prompt injection and jailbreak attempts in real-time. |
|
|