ITmedia AI+
Original published:
It's not "100% safe if you run overseas models on your own server" - Yasushi Amano of Team Mirai points out, emphasizing the necessity of domestic models
AI summary
Editorial policyYasushi Amano of Team Mirai pointed out the risks of intentional performance degradation and data leakage when running overseas open-weight models domestically, emphasizing the need for domestic models.
Key points
- •It was pointed out that overseas open-weight models running domestically could intentionally degrade their own performance.
- •Amano explained that there is a risk of data leakage associated with the use of overseas models.
- •The necessity of developing and using domestic models was emphasized due to these security concerns.