With GenAI Protection, Acronis incorporates generative AI governance into its MSP platform

As artificial intelligence (AI) usage expands, it brings with it a host of risks that existing security measures were not designed to address. To tackle this gap, Acronis has introduced GenAI Protection, an integrated layer within its Cyber Protect Cloud platform aimed at managed service providers (MSPs) responsible for securing small and medium-sized enterprises (SMEs). This solution focuses on three key areas: visibility into unauthorized AI usage, protection of data engaged with AI tools, and prevention against prompt injection techniques.

The uptake of generative AI in organizations hasn’t been executed through controlled deployment but rather spread organically through browsers, SaaS tools, and consumer applications that employees use outside of IT-sanctioned channels. For MSPs tasked with safeguarding SMEs, this trend creates a new category of risk wherein everyday usage poses a greater threat than external intrusions. When an employee submits a contract under negotiation to ChatGPT to extract key clauses, they inadvertently expose sensitive data. Similarly, a prompt injected into an automated workflow can compromise outputs or introduce malicious content into processing chains.

GenAI Protection is seamlessly integrated into the Acronis Cyber Protect Cloud platform, which operates on a distribution model grounded in MSP partnerships. Acronis International GmbH, a Swiss company based in Schaffhausen, navigates legal differentiations from American publishers directly subject to the extraterritorial obligations of the Cloud Act. European customers bound by GDPR and NIS2 will need to address data center localization issues—information on which can be found on the vendor’s website—prior to launching production deployments.

Visibility into Shadow AI: Mapping Before Governance

The first function of GenAI Protection focuses on discovering and monitoring generative AI tools used in client environments, including those adopted outside approved IT channels. The solution identifies active tools, usage trends, and potential exposures, offering actionable reports to support the enforcement of security policies. Unauthorized tool blocking can be activated through URL filtering, leveraging existing EDR/XDR capabilities already present in the Cyber Protect Cloud platform.

This approach addresses a well-recognized structural problem faced by Chief Information Security Officers (CISOs): the inability to govern what isn’t visible. Generative AI tools accessed via browsers typically do not appear in conventional asset inventories and do not trigger alerts in SIEM systems unless detectable data exfiltration occurs. GenAI Protection provides telemetry specific to these use cases, distinct from typical network monitoring flows, and is aimed at the behavioral detection of risky practices before they lead to incidents.

Protection of Sensitive Data in Prompts

The second key function analyzes the content of prompts submitted to generative AI tools and blocks the unauthorized transmission of sensitive data to public or unsanctioned services. Covered categories include personal data, health information, PCI-DSS related payment card data, and any organization-specific confidential information. Blocking occurs prior to submission, ensuring that data does not leave the controlled perimeter.

For organizations subject to GDPR, this capability directly tackles the risk of unintentional violations through the submission of personal data to third-party AI models whose processing conditions may not comply with European regulations. It complements Acronis’s existing Data Loss Prevention (DLP) module, which encompasses over seventy local and network channels for preventing data leakage, including generative AI applications installed on desktops. GenAI Protection manages the browser scope, while Acronis DLP caters to local channels, with both components working in concert.

Blocking Prompt Injection Techniques

The third function addresses prompt injection attacks, an expanding offensive vector as generative AI models are integrated into automated workflows. Prompt injection involves introducing instructions into a processing chain to manipulate the AI model’s behavior, either altering its outputs or extracting sensitive information accessible within the session context. These attacks can originate from the user itself or from external content introduced into the workflow, such as documents, web pages, or automatically retrieved messages.

GenAI Protection detects and blocks these techniques before they can reach the target AI model. For MSPs whose clients deploy AI agents on uncontrolled data, this capability mitigates risks that traditional security tools do not inherently cover. Prompt injection does not trigger network alerts or leave traces in conventional access logs, operating instead at the application level within the language processing layer.

GenAI Protection is available within Acronis Cyber Protect Cloud for managed service providers. Acronis serves over 21,000 MSPs protecting more than 750,000 organizations. This solution integrates into the existing platform without the need for a separate console or independent architecture, thereby reducing integration overhead for providers already equipped. Pricing details have not been disclosed in available sources.

Scroll to Top