In recent years, the concept of AI safety has gained significant traction, evolving into a multifaceted field that extends far beyond the confines of Silicon Valley. When you think about AI safety, do you only envision researchers preventing the emergence of a rogue superintelligence? The reality is much broader. This sector has bifurcated into two major domains: Digital AI Safety, which ensures that large language models and autonomous agents are secure and aligned with human values, and Physical Public Safety, which employs technologies such as computer vision and autonomous hardware to protect lives in the real world. In this article, we’ll explore the key players transforming both facets of AI safety.

Autonomous Surveillance and Defense Technologies

Companies are now integrating AI into physical systems—like drones and sensor towers—to enhance security across various environments, including national borders and critical infrastructures.

  • Anduril Industries: Founded by Palmer Luckey, the mind behind Oculus, Anduril focuses on defense technology. Its flagship software, Lattice, combines data from autonomous systems to detect threats in real-time. Recently, it secured a hefty $192 million contract with the U.S. Army for its rugged mobile hardware.
  • Shield AI: Centered around its AI pilot named Hivemind, Shield AI specializes in autonomous flight in perilous environments, where human pilots may be at risk. They recently integrated their counter-UAS software with the L3Harris VAMPIRE system for improved threat tracking.

    • Innovations in Emergency Response with AI

    When disasters strike, traditional 911 systems can quickly become overwhelmed. Startups are now leveraging AI to streamline emergency response.

  • Carbyne: This platform enhances 911 calls by pulling live video and accurate location data from callers’ smartphones. It utilizes AI to quickly triage emergencies, ensuring first responders receive vital information promptly.
  • Prepared: Designed to work alongside existing 911 systems, Prepared translates non-English calls in real-time and extracts crucial information from chaotic audio, speeding up response times for police and fire departments.

    • Advanced Computer Vision for Crisis Detection

    These platforms continuously monitor video feeds to provide instant alerts, minimizing human response time during crises.

  • ZeroEyes: Founded by military veterans, ZeroEyes integrates its AI into current surveillance systems, allowing for immediate detection of firearms in various settings. They have committed $10 million for R&D to expand into drone-based threat mapping.
  • Pano AI: Focusing on natural disaster detection, Pano AI uses ultra-HD cameras to identify smoke from wildfires almost immediately, thereby aiding utility companies with predictive modeling.
  • Epirus: This company specializes in neutralizing rogue drone swarms using advanced AI targeting systems, safeguarding critical infrastructures and events.

    • Threat Intelligence and Predictive Analytics

    Organizations in this space analyze vast amounts of data to preemptively identify risks.

  • Dataminr: A leader in threat detection, Dataminr processes billions of data points to alert responders and organizations about emerging risks in real-time.
  • Vannevar Labs: This firm focuses on national security, utilizing AI tools to analyze foreign communications, enabling agencies to anticipate conflicts and physical threats.

    • Institutional Safety: The Challenges of AI in Corrections

    When AI is applied in correctional facilities and law enforcement, the complexities increase significantly.

    LEO Technologies (Verus AI)

    LEO Technologies is dedicated to public safety monitoring within correctional settings. Their primary tool, Verus, monitors inmate communications, aiming to enhance safety. However, independent assessments have revealed issues like high false-positive rates, raising questions about the reliability of such AI systems in chaotic environments.

    ViaPath Technologies (GTL)

    Unlike specialized software firms, ViaPath Technologies controls the core infrastructure for corrections. They have contracts with numerous facilities and offer AI-driven platforms that monitor inmate communications. Through extensive data processing, they flag potential risks, but their dominance raises concerns about surveillance and civil liberties.

    The Expanding Landscape of AI Safety

    As AI safety continues to evolve, it encompasses a range of activities—from aligning code to using sensors for real-world monitoring. Whether it’s a company like Anthropic focusing on ethical AI training or ZeroEyes providing real-time threat assessments, the landscape is more diverse than ever. Understanding these advancements is crucial as we navigate the complexities of AI in our everyday lives.

    RELATED ARTICLESMORE FROM AUTHOR