SAP Completes Prior Labs Acquisition

SAP has completed the acquisition of Prior Labs, the pioneer of Tabular Foundation Models (TFMs). The enterprise software giant is committing to investing more than €1 billion over the next four years to scale Prior Labs into a globally leading frontier AI lab for the structured data that underpins the world’s businesses.

Independence Maintained

Prior Labs will continue to operate as an independent entity within SAP, preserving the research lab’s autonomy while benefiting from SAP’s resources and reach.

Agentic AI Security Shifts to Structural Defenses

July 2026 marked a critical turning point for Agentic AI security, shifting focus from theoretical vulnerabilities to structural, systemic defenses. High-profile frameworks from Google DeepMind and Anthropic underscore that AI agents must be treated as potential insider threats, requiring robust, cryptographically verifiable guardrails, strictly scoped identities, and comprehensive runtime monitoring.

Emerging Threat Landscape

Research has revealed concrete exploitation risks across multiple agent platforms:

  • Researchers found an 85% exploitation success rate when planting attacker instructions in fake Sentry error reports executed by AI coding agents.
  • Microsoft details the AutoJack exploit chain against AutoGen Studio, which weaponizes an agent’s browsing to reach a privileged localhost service and execute code on the host with no user interaction.
  • A systematic study of multi-agent systems like MetaGPT and CrewAI found that the IMA attack achieves 89% success, highlighting that multi-agent collaboration amplifies harm significantly over single-agent baselines.

Source: SAP News Center