Shingo Takagi, Marketing/BizDev A10Networks, Japan (@shingo_a10) header
Shingo Takagi, Marketing/BizDev A10Networks, Japan (@shingo_a10) profile picture

General Manager, Business Development at A10 Networks KK, RELIABLE SECURITY ALWAYS | BtoB marketer, 22 years’ in Networking and security area

Tokyo, Japan
Joined December 2011
Photo by Shingo Takagi, Marketing/BizDev A10Networks, Japan (@shingo_a10)Photo by Shingo Takagi, Marketing/BizDev A10Networks, Japan (@shingo_a10)Photo by Shingo Takagi, Marketing/BizDev A10Networks, Japan (@shingo_a10)Photo by Shingo Takagi, Marketing/BizDev A10Networks, Japan (@shingo_a10)Photo by Shingo Takagi, Marketing/BizDev A10Networks, Japan (@shingo_a10)Photo by Shingo Takagi, Marketing/BizDev A10Networks, Japan (@shingo_a10)Photo by Shingo Takagi, Marketing/BizDev A10Networks, Japan (@shingo_a10)Photo by Shingo Takagi, Marketing/BizDev A10Networks, Japan (@shingo_a10)Photo by Shingo Takagi, Marketing/BizDev A10Networks, Japan (@shingo_a10)Photo by Shingo Takagi, Marketing/BizDev A10Networks, Japan (@shingo_a10)
When a PHP vulnerability emerged on October 21, ThreatX had a blocking rule live 14 minutes after public disclosure. Attackers don't wait for patch windows. Neither should your defense. sprou.tt/1thCsrUjkWr
16
As agents act, your attack surface expands. AI security goes beyond the prompt. Teams need to know every tool and verify every change. Test before launch and defend at runtime. Learn more: bit.ly/3WduVib
29
Single-turn testing only tells you if one command works. Real attackers adapt, build context, and exploit your AI's logic over multiple turns. Jamison Utter and Priyanka Mullan of @A10Networks explain why AI red teaming has to go multi-turn 👇 sprou.tt/1PZi2yeIYrC
23
Securing your AI means testing beyond the model. Jamison Utter and Priyanka Mullan of @A10Networks explain how red teaming LLM logic, context, and decision boundaries helps prevent unauthorized access and data exposure. sprou.tt/1sVGkvPVhTR
2
73
Here's what an AI-native security layer should actually do. bit.ly/4cgMzY3
11
A model can refuse an unsafe request and still sit inside an unsafe application. Securing AI agents takes five controls around the model: guardrails, tool access, discovery, kill switches, and red teaming. sprou.tt/15WbSUvIvJd
7
Deploying AI in production means solving three problems at once.
15
As MCP connects agents to tools and data, risk shifts to the action sequences they follow. Prompt injection, tool poisoning, and multi-step data exposure need new defenses. Traditional AppSec isn't enough. Govern the agent loop. sprou.tt/1lTn9pRxDoa
1
61
Is your AI gateway doing enough? Learn how next-gen gateways centralize governance, route requests to the right model, and control token spend across every team and LLM provider. Download the white paper: bit.ly/4zn6VI8
1
32
AI attacks are now autonomous, adaptive, and multi-step. Static prompts won't cut it. TrojAI Detect's agent-led red teaming simulates multi-turn attacks, adapts to responses, and maps findings to OWASP, MITRE & NIST. sprou.tt/1EQ3cjFd9hk
23
Legacy tools can’t handle today’s expanding attack surface. AI-driven security platforms simplify defense by combining detection, prevention & response—all in one agile system. Learn how AI offers unprecedented leverage. #AI sprou.tt/1COXZRSQ9Bp
1
18
Agentic AI is moving from simple chat to autonomous action. But how do we secure AI that can call its own APIs? @A10networks Gary Wang explains the role of Model Context Protocol (MCP) and why an AI Firewall is the essential guardrail for modern workflows. sprou.tt/1dG6xX1zXzB
2
70
What is model abliteration? Model abliteration enables deeper AI safety testing by creating cooperative research models. Read this blog post to learn how A10 Networks' TrojAI solution uses abliterated LLMs to uncover risks and harden security. sprou.tt/1SRX2oIwLSA
22
Shingo Takagi, Marketing/BizDev A10Networks, Japan retweeted
【タイムズカー約660万件情報漏えい】 news.yahoo.co.jp/pickup/6596…
447
5,981
5,412
16,317
81,604,024
Ever wonder how AI agents and chatbots actually talk to LLMs? Our on-demand webinar breaks down the API calls behind the interface: system prompts, context windows, and why these mechanics matter for AI red teaming. #AI #Security sprou.tt/1bSj4Rxp3xi
40
As AI systems move from experimental tools to operational infrastructure, the nature of risk has changed dramatically. What was once a matter of model accuracy is now a broader challenge of system integrity, adversarial resilience, and real-world impact. sprou.tt/1b3PszmogxI
14
Prompt guardrails can see an AI agent's tool call, but not its intent. As agents become autonomous via MCP, security must cover the whole workflow, not just the prompt. sprou.tt/1rtxMQm4cwK
31
Agentic AI and MCP are reshaping workflows, but they also expand the attack surface for critical APIs. Securing the AI era requires moving beyond static defense to adaptive API protection and AI-specific firewalls. sprou.tt/1p7jKplL3nF
30
Upwards of 80% of tech companies have adopted GenAI, but their infrastructure isn't always keeping pace. A10's recent AI infrastructure survey highlights these constraints and growing concerns about AI security.  sprou.tt/1kRqAQa8Jxr
1
22
Here's what an AI-native security layer should actually do. bit.ly/4cgMzY3
17