AIエージェントの開発・提供におけるセキュリティリスクの論点と対策

2026年9月2日（水）開催「Product Security Square」に登壇した梅内のスライド資料です。
https://flatt.connpass.com/event/393895/

Transcript

1. 2026/09/02 GMO Flatt Security
2. Umeuchi Tsubasa 02 / 23
3. 01 02 03 03 / 23
4. CHAPTER 01
5. Takumi byGMO shisho.dev/docs/ja/t/ 05 / 23
6. Takumi Guard Takumi Runner Takumi Images 06 / 23
7. 07 / 23
8. OpenAI "Designing AI agents to resist prompt injection" Anthropic "How we contain Claude across products" 08 / 23
9. CHAPTER 02
10. ※ 10 / 23
11. Takumi API shisho.dev/docs/ja/ 11 / 23
12. AI Agent SaaS SRE NEXT 2026 speakerdeck.com/flatt_security 12 / 23
13. AI Agent SaaS SRE NEXT 2026 speakerdeck.com/flatt_security 13 / 23
14. 14 / 23
15. Takumi shisho.dev/docs/ja/t/assessment/architecture/blackbox 15 / 23
16. Takumi shisho.dev/docs/ja/t/assessment/architecture/blackbox 16 / 23
17. CWE-400 Uncontrolled Resource Consumption 17 / 23
18. Takumi API shisho.dev/docs/ja/ 18 / 23
19. 19 / 23
20. CHAPTER 03
21. 01 02 03 04 05 21 / 23
22. ASI01 ASI02 ASI03 ASI04 ASI05 ASI06 ASI07 ASI08 ASI09 ASI10 OWASP Top 10 for Agentic Applications 2026 genai.owasp.org 22 / 23
23. Our Mission