Navigating the M-25-22 AI Acquisition Mandate

The Shift to M-25-22

In April 2025, the Office of Management and Budget (OMB) released Memorandum M-25-22, which rescinded and replaced the previous guidance, M-24-18 [28]. For federal contractors, this shift represents a refinement in how agencies are expected to procure and manage Artificial Intelligence (AI) systems [28]. While the core focus remains on 'responsible' acquisition, M-25-22 emphasizes efficiency and lifecycle cost management, forcing contractors to be more transparent about the long-term operational costs of their AI solutions [26, 28].

Lifecycle Cost Transparency

One of the most significant impacts of M-25-22 is the requirement for agencies to conduct rigorous financial planning for the entire lifecycle of an AI system [26]. Contractors should expect RFPs to demand detailed pricing transparency that covers not just the initial development, but also post-award oversight, maintenance, and corrective actions [26].

Agencies are now instructed to require vendors to provide pricing transparency across the total lifecycle of AI design, development, and duration of use [26]. This includes specific requirements for cloud-based AI services, such as the avoidance of egress fees, which can often inflate the total cost of ownership [26]. Capture managers should prepare their pricing volumes to address these transparency requirements explicitly, as agencies are under pressure to demonstrate fiscal responsibility in their AI investments [26].

Risk Management and Performance Testing

Under the new guidance, contracts for AI systems must detail the examination, testing, and validation procedures for which the vendor is responsible [26]. This is particularly critical for 'rights-impacting' or 'safety-impacting' AI [22, 26]. Contractors must be prepared to provide results of performance testing for algorithmic discrimination and bias [26].

Furthermore, agencies are encouraged to leverage the AI Safety Institute’s testing protocols [26]. Contractors should anticipate that their AI models will be subject to third-party evaluations and that they will need to provide documentation that aligns with these federal standards [26].

Strategic Positioning for AI Pursuits

To win in this environment, contractors must move beyond generic AI capabilities. Your proposal strategy should include:

• Data Governance Plans: Clearly articulate how you mitigate risks like data poisoning or data leakage when training models with agency data [26].
• Environmental Impact Statements: When proposing enterprise-wide generative AI, be prepared to discuss the environmental impact of training and operating your models [26].
• Licensing Transparency: Ensure your licensing terms are clear and do not restrict the agency’s ability to use the acquired AI system in reasonably expected ways [26].

By aligning your proposal content with the specific requirements of M-25-22, you demonstrate that your firm is not just an AI provider, but a partner in the government’s mission to deploy AI safely and efficiently [26, 28].