How to Secure Proposal Assets in the Age of Agentic AI

The world witnessed a large-scale cyber-espionage campaign powered almost entirely by Agentic AI in September 2025. According to Anthropic’s Threat Report, a Chinese state-sponsored group used Claude’s coding tools to automatically scan systems, breach networks, and steal data from 30 organizations.

This incident didn’t just make headlines—it delivered a clear warning to every industry using AI, including federal contractors. For proposal managers handling sensitive capture intelligence, pricing, corporate information, partner data, and win strategies, the message is unmistakable: Agentic AI is powerful and transformative, but it can also be dangerous if not secured.

This blog explains how proposal professionals can use Agentic AI to expedite proposal development and, just as importantly, how to work with their Corporate Security Officers (CSOs) to protect corporate proposal assets from unwanted intrusions.

How Proposal Professionals Can Use Agentic AI to Develop Proposals

Agentic AI refers to artificial intelligence systems that can reason, plan, take actions, and iterate across multiple steps autonomously, without needing continuous human prompts. Here are three examples of how Agentic AI can help transform proposals:

• Requirements Analysis: Agentic AI automatically interprets RFP sections, extracts requirements, and maps them into outlines or compliance matrices. This provides proposal teams with an immediate head start but also requires human verification.
• Proposal Planning: Agentic AI can help sequence tasks, track deadlines, notify writers of milestones, manage version control, and keep teams on track.
• Content Development: Agentic AI can produce draft text, but proposal managers must edit the text to improve accuracy, relevance, and quality.

Examples of Agenctic AI tools include:

• Microsoft Copilot Agents: These tools let you build task-specific AI assistants that work inside Microsoft 365 apps like Teams, Outlook, and SharePoint.
• OpenAI Operator: This tool from OpenAI gives AI models access to external tools and Application Programming Interfaces (APIs), enabling them to complete multi-step tasks independently.
• CrewAI: This framework lets multiple AI agents manage and complete complex workflows.

• IBM watsonx Orchestrate: This platformconnects enterprise systems and automates workflow using AI-driven agents.
• Amazon Quick Suite (AWS): This set of agentic AI tools supports business tasks, research, automation, and data analysis.

How to Secure Proposal Assets Against Agentic AI Tools

While Agentic AI offers proposal professionals numerous advantages, it can also be dangerous if not appropriately secured. Lohfeld Consulting recommends that proposal professionals work with their CSOs to identify how Agentic AI could be used to target or access their proposal assets, and to develop primary and backup plans to safeguard those assets in the event of an attack. To build your proposal protection strategy, consider asking your CSO the following questions:

• How can we establish clear AI use policies? Identify which tools are approved, what content may or may not be uploaded, when human intervention is required, how AI-generated content is stored, and how long it is retained.
• How can we build a secure proposal knowledge library? Discuss how you can verify your library has role-based permissions, version control, expiration rules, and encryption.
• How can we use enterprise AI tools while also limiting reliance on public AI models? Discuss how you can guarantee that public AI models will not train on your data. Discuss how to implement data encryption techniques, System and Organization Controls (SOC) 2/ Federal Risk and Authorization Management Program (FedRAMP)-level controls, and audit logs of agent actions.
• How can we set up a protected “AI Proposal Zone”? Discuss how to limit AI role-based access to proposal assets. This includes limiting access by job role, defining which AI tools may be used, locking down folders containing pricing or corporate financials, and restricting teaming agreements to only those who need them. By limiting who can interact with these assets, organizations create a secure environment that reduces the risk of data leakage or unintended actions.

Conclusion: Agentic AI Offers Speed, but Only If You Control the Risk

The Agentic AI-driven cyberattack in September 2025 showed how quickly agentic systems can be weaponized. For proposal managers, it reinforces a simple truth: you must pair AI innovation with disciplined security. By adopting secure enterprise tools, establishing guardrails, and keeping humans in the loop, proposal teams can harness Agentic AI without exposing sensitive assets. If your team needs training in how to use AI/GenAI—and use it responsibly—consider taking Lohfeld Consulting Group’s Generative AI for Proposal Professionals class or hiring our consultants, who are trained in the use of GenAI, to help you on your next proposal. Contact us for more information.

Relevant Information

• How AI Tools Can Transform Proposal Management Now

• How to Maximize AI and GenAI Implementation Now

• Automated Tools Powering Growth Beyond AI Now

By Brenda Crist, Vice President at Lohfeld Consulting Group, MPA, CPP APMP Fellow

Lohfeld Consulting Group has proven results specializing in helping companies create winning captures and proposals. As the premier capture and proposal services consulting firm focused exclusively on government markets, we provide expert assistance to government contractors in Capture Planning and Strategy, Proposal Management and Writing, Capture and Proposal Process and Infrastructure, and Training. In the last 3 years, we’ve supported over 550 proposals winning more than $170B for our clients—including the Top 10 government contractors. Lohfeld Consulting Group is your “go-to” capture and proposal source! Start winning by contacting us at www.lohfeldconsulting.com and join us on LinkedIn, Facebook, and YouTube^(TM).