Loading story…
The UK National Cyber Security Centre (NCSC) warned that “shadow AI”—employees using AI tools outside approved organisational systems—can expose sensitive corporate, proprietary and customer data while depriving security teams of visibility and control. Citing Microsoft research, it said 71% of UK employees have used AI tools not approved by their employer; information submitted to consumer AI services may be retained, stored or used to improve those services, creating breach, intellectual-property and regulatory-compliance risks.
The NCSC also cautioned that vulnerable, poorly governed or compromised AI agents could provide attackers with access to the data, services and privileges available to those agents. It advised organisations to reduce rather than expect to eliminate shadow-AI use by fostering open communication, understanding employee requirements, establishing secure-use guardrails, offering usable approved alternatives, and carefully governing agentic AI integrations.
Track how attackers are adapting to this technology.
1 event from the most recent confirmed update back to the earliest known activity.
The UK National Cyber Security Centre published guidance warning that employee use of AI tools outside approved systems can expose corporate and customer data, reduce organisational visibility and control, and create breach, intellectual-property, and compliance risks. It also warned that exploitation of vulnerable AI agents could give attackers access to the data, services, and privileges available to those agents, and recommended secure-use guardrails, approved alternatives, and an open security culture.
Follow how adversaries are adapting to this technology, and where it touches your stack today.
3 references tracked. Mallory keeps watching after this page renders.
itpro.com
Open sourceinfosecurity-magazine.com
Open sourcencsc.gov.uk
Open sourceMap indicators from this story to your assets and identify affected systems in minutes.
Every observed campaign, victim, and pivot linked to actors named in this story.
Malware, exploits, and IOCs connected to the activity described here.
YARA, Sigma, and Snort rules deployed to your SIEM as soon as they’re published.
Get matching new stories delivered to your team as they break — not the next morning.
Ask questions about this story and take action on the answers.
Other stories Mallory is tracking that overlap on entities, sources, or topic labels.
Organizations are facing a growing challenge as employees increasingly adopt AI tools and agents without formal IT approval, a phenomenon known as shadow AI. This unsanctioned use of AI—ranging from chatbots and large language models to low-code agents—enables employees to automate workflows and make decisions outside traditional governance structures. The lack of oversight and visibility into these autonomous systems exposes enterprises to significant risks, as sensitive data may be processed or shared through unvetted platforms, and decisions may be influenced by tools that operate beyond established security controls. Recent research highlights that 73% of employees use AI for work, yet over a third do not consistently follow company policies, and many are unaware of existing guidelines. About 27% admit to using unapproved AI tools, often browser-based and free, making them difficult for IT to monitor. This shadow AI trend compounds the broader issue of shadow IT and SaaS sprawl, where employees bypass official channels to access tools that better meet their needs. Security teams are advised to shift from outright bans to strategies focused on discovery, communication, and oversight to manage these risks effectively.
Jul 1, 2026
Security leaders are warning that rapid adoption of AI tools—often outside formal governance—creates expanding blind spots and increases the likelihood of **data leakage** and operational incidents. A webcast discussion framed “**Shadow AIT**” as the AI-era evolution of shadow IT, highlighting that AI capabilities are frequently embedded in everyday SaaS features and browser extensions, making it difficult for organizations to accurately inventory where AI is in use and what data is being shared. The panel cited a cautionary example involving *Replit* where insufficient controls around an AI agent reportedly contributed to a production database deletion, underscoring that agentic workflows can translate governance gaps into real outages. Separately, reporting on *Google Vertex AI* raised concerns that **permissions and access control design** in AI platforms can amplify **insider-risk** scenarios if roles, entitlements, and auditability are not tightly managed—particularly where AI services can access or act on sensitive datasets. Commentary-style content also broadly discusses “cognitive AI” and future-facing architectures, but without tying to a specific incident or disclosure; the actionable takeaway across the relevant items is to treat AI enablement as an identity, data-governance, and monitoring problem (inventory AI usage, constrain permissions, and instrument logging) rather than a purely productivity tooling decision.
Jun 29, 2026
Security researchers, vendors, and practitioners warned that **AI agents and assistant tools are creating major visibility and governance gaps** across endpoints, SaaS platforms, and cloud infrastructure. At Black Hat and BSides Las Vegas, speakers said traditional human-centric security models do not fit autonomous systems that act continuously through APIs, combine trust with changing context, and can quickly exceed intended authority. Recent reporting and commentary highlighted recurring failures such as treating agents like ordinary identities, granting broad permissions for convenience, underestimating API access, and missing “shadow AI” deployments, while guidance from the Cloud Security Alliance and others stressed **runtime authorization, per-action controls, requester-aware logging, least privilege, and reduced blast radius** for agentic systems. New examples showed how those risks are surfacing in practice. Imperva released **ShadowAI-Watch**, an open-source host monitoring tool that correlates connections to LLM providers with process trees, commands, file access, and network activity, and demonstrated detection of a browser agent manipulated by indirect prompt injection into attempting session-data exfiltration. SecuritySenses warned that bot-less AI meeting notetakers can evade participant-list visibility while retaining broad OAuth access to calendars, drives, and future meetings, and a Reddit post alleged Microsoft automatically enabled **OpenAI as a Microsoft 365 Copilot subprocessor** for some tenants, raising compliance concerns. SecurityWeek also reported that AI accelerators and “neo-clouds” introduce infrastructure blind spots beyond traditional CPU telemetry, increasing the risk of cross-tenant leakage, model theft or poisoning, and covert abuse unless organizations extend monitoring and policy controls into these emerging AI execution environments.
Aug 17, 2026