Cantivia

Raising $25,000active

Cantivia detects deepfakes and disinformation, and simulates how synthetic media spreads, so platforms, newsrooms and fraud teams know what's fake, who made it, and how far it'll travel.

Project Details

Cantivia is an AI-security company tackling synthetic media end to end. Praetorian detects deepfakes across image, audio and video and fingerprints which model generated them, using DCT frequency-domain analysis on a custom backbone. Our second tool flags disinformation at 99% accuracy, built by retraining layers of a 30B model, which we use to verify the truth. SEDPNR, a 10,000-agent simulated population built from survey data on 1,000+ real people, models how a piece of media spreads amongst an artificial population. We serve platforms, newsrooms and fraud/trust-and-safety teams who need to know what's fake, who made it, and how far it will travel. I (Ashkan) lead Praetorian; my co-founders Feargal and Oisin lead the disinformation classifier/SEDPNR and operations/go-to-market respectively. We've been building together since January 2026 and met through school 3 years ago.

Theory of Impact

As generative AI gets cheaper and better, most of what people see, hear and read online will be synthetic, so verifying what's real, who made it, and how it spreads becomes essential infrastructure. Detection hasn't kept up: it's slow, single-modality, offers no attribution, and no spread modelling. Cantivia builds that trust layer across all three, giving platforms, newsrooms and fraud teams a way to catch fakes, trace them to source, and anticipate their reach before they cause harm. This strengthens the human/machine boundary in an information environment that is degrading as synthetic content scales.

People

Ashkan Samali

Team Member

Grants Received

Add grant

Emergent Ventures Grant

from Emergent Ventures (Tyler Cowen)

$20,000

Funding Asks

grantmaking.ai Launch Round

Applied

Minimum

$25,000

Ideal

$100,000

How the money will be spent

Minimum ($25,000) covers the final engineering push plus baseline infra for about 6 months:

• Compute for model training/inference (Praetorian detection + disinformation classifier retraining): $10,000

• Cloud hosting & API serving (AWS/GCP, storage, bandwidth): $4,000

• SEDPNR simulation compute (running the 10,000-agent population): $3,000

• Tooling & software subscriptions (GitHub/CI, monitoring, vector DB, data labeling, third-party LLM API calls): $3,000

• Incorporation & legal (company formation in Ireland or Delaware, basic contract review): $3,000

• Contingency (domains, business software, insurance): $2,000

Ideal ($100,000) adds the above plus runway to hire and scale for the September demos:

• First engineer, part-time/contract for about 6 months: $30,000

• Additional compute to scale detection models and SEDPNR to larger populations: $20,000

• Security/compliance tooling and hardening ahead of enterprise customer demos: $8,000

• Expanded cloud storage & data pipeline for the three customer demos: $7,000

• Go-to-market costs (demo materials, customer travel/meetings): $5,000

• Contingency/buffer: $5,000

These are our current best estimates rather than vendor quotes, and we'll refine them as we finalise which cloud/compute providers we use.

Discussion

No comments yet. Be the first to share your thoughts.