Neural Defend

Neural Defend is an AI cybersecurity company that develops deepfake detection technology for video, audio, images and documents. The platform is positioned as enterprise-grade and is described as identifying AI-generated manipulations across four media types: video (frame-level analysis, blink and micro-expression tracking, temporal consistency and compression artifact scanning), audio (spectrogram pattern matching, voice cloning signature detection, background noise consistency), images (AI-generation fingerprinting, face-swap boundary detection, metadata integrity, multi-face and lighting/shadow consistency analysis) and documents (manipulation detection and metadata verification).

The company states its proprietary models detect deepfakes across video, image, audio and real-time streams, and that capital raised in 2025 is being directed toward agentic AI capabilities for real-time, multimodal detection alongside R&D and product development. Detection is offered through three deployment modes: a REST API with JSON responses and OAuth2 authentication, native client-side SDKs for mobile and web, and an on-premise Docker deployment intended for air-gapped or data-sovereignty-constrained environments. Integration is described as a three-step process covering authentication, media analysis and enforcement based on returned risk scores. The company reports integrations with Zoom, Microsoft Teams, Google Meet and Webex for real-time protection during video calls.

Neural Defend was founded in 2024 by Piyush Verma (CEO), Sivashankar Selvarajan (CTO) and Sumit Singh (Chief Data Officer), and describes itself as born at MIT and supported by the MIT Venture Mentoring Service. As of March 2025 reporting, the company was pre-revenue and running pilot projects in New York and Singapore.

Founding story

The founders describe leaving established careers to address authenticity in synthetic media: CTO Siva Shankar left an AI industry role after work at Intel, IBM (via UST Global), Axis Securities and Cisco Systems, and CDO Sumit Singh moved from a data role tied to Invesco through Publicis Sapient. CEO Piyush Verma is a two-time founder and former researcher at MIT, Harvard and Tsinghua University who previously founded Manush Labs and the MIT India Initiative. The company describes itself as born at MIT and backed by its Venture Mentoring Service.

Business model

Neural Defend sells deepfake detection as a technology platform to organizations, offered via REST API, native SDKs for mobile and web, and on-premise Docker deployment behind a customer firewall. As of March 2025 the company was reported to be pre-revenue, engaging global enterprises, fintech firms and financial institutions for product validation and market expansion.

Traction

As of March 2025 the company was pre-revenue, with pilot projects underway in New York and Singapore and engagement with global enterprises, fintech companies and financial institutions for product validation. The company claims state-of-the-art benchmarks for its technology, and its website cites more than 10 million files analyzed, 100% detection accuracy, sub-one-second processing time and a 99.9% uptime SLA.

Latest developments

In March 2025 the company announced a pre-seed round of over $600,000 led by Inflection Point Ventures with MIT SBXI, Techstars San Francisco and Soonicorn Ventures participating, earmarked for agentic AI capabilities, R&D and product development. Its website cites 2025 recognitions from the Data Security Council of India and selection among the top 12 Bharat AI startups by NPCI and NVIDIA at the Global Fintech Festival 2025.