← Voltar ao feed

Netskope: Manufacturing AI use jumps to 67% of users as shadow AI, MCP activity and malicious AI links raise data exposure risk

1 fonte(s) · 1 atualização(ões) · publicado 16/11/2026 11:02 · coletado há 1 dia

Netskope Threat Labs published a November 2026 report on AI adoption in the manufacturing sector. It is based on anonymized Netskope One telemetry from 1 June 2025 to 31 August 2026. Active AI users rose from 40% to 67%, and organization-managed AI use rose from 35% to 70%. Personal AI account use fell from 75% to 55%, but the share of users switching between personal and enterprise accounts grew from 9% to 25%. The report describes a risk trend, not a specific incident, and names no CVEs, threat actors or malware.

Detalhes técnicos

  • Microsoft Copilot is the most widely adopted AI application, used by 90% of organizations. Claude Platform follows at 89% and ChatGPT at 76%.
  • 96% of employees use applications with embedded AI features. 90% interact with systems that use customer or user data for model training.
  • Regulated data accounts for 57% of AI-related data policy violations. Passwords and keys account for 19%, intellectual property for 16% and source code for 9%.
  • Non-human entities connecting to remote Model Context Protocol (MCP) servers rose 490%, and MCP events rose 600%.
  • Malicious AI link interactions rose from July 2025 and reached 150 per 100,000 users in October 2025. They rose again from February 2026 and reached 200 per 100,000 users in March 2026. AI-themed lures spiked in February and June 2026. Netskope cites AI engine optimization (AIEO) and advertising inside AI applications as emerging channels for malicious content.
  • The most frequently blocked AI applications are Particular Audience (45% of organizations), Emergent (39%) and DeepSeek (36%).

Remediação

  • Inspect all HTTP and HTTPS downloads.
  • Block applications with no legitimate business purpose.
  • Apply DLP policies to AI and personal app destinations.
  • Use Remote Browser Isolation for higher-risk site categories.
  • Gain visibility and real-time control over managed and unmanaged AI apps and API interactions.
  • Treat links returned by AI applications like links in search results and email.
  • Red-team private AI deployments.
  • Speed up the approval process for new AI tools to reduce shadow AI.

Linha do tempo

  1. industrialcyber.co ↗ · pub. 16/11/2026 11:02 · coletado 16/11 12:01 primeiro registro

    Netskope: Manufacturing AI use jumps to 67% of users as shadow AI, MCP activity and malicious AI links raise data exposure risk

    ler a análise desta fonte

    Netskope Threat Labs published a November 2026 report on AI adoption in the manufacturing sector. It is based on anonymized Netskope One telemetry from 1 June 2025 to 31 August 2026. Active AI users rose from 40% to 67%, and organization-managed AI use rose from 35% to 70%. Personal AI account use fell from 75% to 55%, but the share of users switching between personal and enterprise accounts grew from 9% to 25%. The report describes a risk trend, not a specific incident, and names no CVEs, threat actors or malware.

    Technical Details

    • Microsoft Copilot is the most widely adopted AI application, used by 90% of organizations. Claude Platform follows at 89% and ChatGPT at 76%.
    • 96% of employees use applications with embedded AI features. 90% interact with systems that use customer or user data for model training.
    • Regulated data accounts for 57% of AI-related data policy violations. Passwords and keys account for 19%, intellectual property for 16% and source code for 9%.
    • Non-human entities connecting to remote Model Context Protocol (MCP) servers rose 490%, and MCP events rose 600%.
    • Malicious AI link interactions rose from July 2025 and reached 150 per 100,000 users in October 2025. They rose again from February 2026 and reached 200 per 100,000 users in March 2026. AI-themed lures spiked in February and June 2026. Netskope cites AI engine optimization (AIEO) and advertising inside AI applications as emerging channels for malicious content.
    • The most frequently blocked AI applications are Particular Audience (45% of organizations), Emergent (39%) and DeepSeek (36%).

    Remediation

    • Inspect all HTTP and HTTPS downloads.
    • Block applications with no legitimate business purpose.
    • Apply DLP policies to AI and personal app destinations.
    • Use Remote Browser Isolation for higher-risk site categories.
    • Gain visibility and real-time control over managed and unmanaged AI apps and API interactions.
    • Treat links returned by AI applications like links in search results and email.
    • Red-team private AI deployments.
    • Speed up the approval process for new AI tools to reduce shadow AI.