AI 稽核方法論

# 核心概念示範
from openai import OpenAI
 
client = OpenAI()
 
def demonstrate_concept(system_prompt: str, user_input: str) -> str:
    """示範基本行為樣態。"""
    response = client.chat.completions.create(
        model="gpt-4o",
        messages=[
            {"role": "system", "content": system_prompt},
            {"role": "user", "content": user_input},
        ],
        temperature=0.0,
    )
    return response.choices[0].message.content
 
# 基線行為
baseline = demonstrate_concept(
    system_prompt="You are a helpful assistant that only discusses cooking.",
    user_input="What is the capital of France?",
)
print(f"Baseline: {baseline}")

import json
from typing import Optional
 
class TechniqueFramework:
    """用於實作與測試所述技術的框架。"""
 
    def __init__(self, target_config: dict):
        self.config = target_config
        self.results = []
 
    def prepare_payload(self, objective: str, constraints: dict) -> str:
        """根據目標與目標系統限制準備攻擊載荷。"""
        # 依目標防禦姿態調整載荷
        payload = self._base_payload(objective)
 
        if constraints.get("input_classifier"):
            payload = self._apply_obfuscation(payload)
 
        if constraints.get("output_filter"):
            payload = self._add_extraction_channel(payload)
 
        return payload
 
    def execute(self, payload: str) -> dict:
        """執行技術並蒐集結果。"""
        response = self._send_to_target(payload)
        success = self._evaluate_response(response)
 
        result = {
            "payload_hash": hash(payload),
            "success": success,
            "response_length": len(str(response)),
        }
        self.results.append(result)
        return result
 
    def report(self) -> dict:
        """生成所有執行結果的彙總報告。"""
        total = len(self.results)
        successes = sum(1 for r in self.results if r["success"])
        return {
            "total_attempts": total,
            "successes": successes,
            "success_rate": successes / total if total > 0 else 0,
        }

from dataclasses import dataclass
from typing import Optional
import time
 
@dataclass
class SecurityGateway:
    """用於保護 LLM 應用存取的閘道器樣式。"""
 
    input_classifier: object  # 基於 ML 的輸入分類器
    output_filter: object     # 輸出內容過濾器
    rate_limiter: object      # 速率限制服務
    audit_logger: object      # 稽核軌跡記錄器
 
    def process_request(self, user_id: str, message: str, session_id: str) -> dict:
        """透過所有安全層級處理請求。"""
        request_id = self._generate_request_id()
 
        # 第一層:速率限制
        if not self.rate_limiter.allow(user_id):
            self.audit_logger.log(request_id, "rate_limited", user_id)
            return {"error": "Rate limit exceeded", "retry_after": 60}
 
        # 第二層:輸入分類
        classification = self.input_classifier.classify(message)
        if classification.is_adversarial:
            self.audit_logger.log(
                request_id, "input_blocked",
                user_id, classification.category
            )
            return {"error": "Request could not be processed"}
 
        # 第三層:LLM 處理
        response = self._call_llm(message, session_id)
 
        # 第四層:輸出過濾
        filtered = self.output_filter.filter(response)
        if filtered.was_modified:
            self.audit_logger.log(
                request_id, "output_filtered",
                user_id, filtered.reason
            )
 
        # 第五層:稽核記錄
        self.audit_logger.log(
            request_id, "completed",
            user_id, len(message), len(filtered.content)
        )
 
        return {"response": filtered.content}