If I'm understanding the question correctly, I would generally recommend continuing to test the 200 even if you've had 8 failures in the first 50.
Once you've failed the lot, I'm assuming someone is going to conduct an investigation (either internally or at your supplier, if it's a purchased part). The person conducting the investigation is going to need to define the problem and assess the risk and impact. If you haven't finished your sampling, it could be hard to impossible to understand the extent of the issue. If you stop inspection, you might miss the quantity of defects or the range of defect types.
For example, compare two scenarios: You test the other 150 and have 0 additional failures, and you test the other 150 and have 24 additional failures (i.e., about 8 for every 50). If 15%+ of the batch is defective, it points to a much bigger failure that needs immediate correction than being a hair over AQL.
If you don't finish your sampling, you also might miss another issue with the product. For example, let's say these are a molded plastic component you purchased from a vendor, and the inspection is cosmetically inspecting 50 samples from the beginning of the run, 50 samples from the end of the run, and 100 samples pulled randomly in-between. Let's say you've only checked 50 samples from the beginning / middle of the run and found excess flashing around the edges. If you don't check the parts from the end of the run, you might miss that there was discoloration, malformed parts, or embedded particulate that didn't show up until the end of the run.
Of course, this depends somewhat on the time / effort / expense of testing each unit. The company I work for has a 100% inspection for certain characteristics. The general rule of thumb is that if the first ~20% of a batch all fails, stop inspection and escalate to a nonconformance. It's just not worth the equipment and operator time to keep inspecting something we're reasonably certain is going to fail anyway.
