Deetech vs Sensity AI: Which Is Better for Insurance Claims?

Sensity AI is one of the longest-running deepfake detection companies in the market. With over eight years of experience, forensic-grade analysis, and strong relationships with law enforcement agencies across Europe, they’ve earned a reputation for technical rigour.

Deetech is purpose-built for insurance claims verification, designed to detect AI-generated and manipulated media at the scale and speed that insurance operations demand.

Both platforms are technically capable. The question for insurance buyers is which one fits the operational reality of a claims department.

Sensity AI: The Forensic Specialist

Sensity AI (formerly Deeptrace) has been operating in the deepfake detection space since approximately 2018, making them one of the earliest dedicated players. Their heritage is in forensic analysis and threat intelligence.

Strengths:

Forensic-grade accuracy. Sensity reports 98% accuracy on public benchmark datasets, a figure that reflects genuine technical capability across controlled test conditions.
Court-ready reports. Their forensic analysis produces detailed technical reports designed for evidentiary use, particularly in law enforcement investigations.
Eight-plus years of experience. Sensity has been tracking the evolution of generative AI since before most organizations recognized the threat. Their institutional knowledge of deepfake techniques is substantial.
Threat intelligence platform. Beyond detection, Sensity monitors deepfake trends, tracks new generation techniques, and provides intelligence on emerging threats.
Law enforcement relationships. Strong track record working with European police forces, Interpol, and government agencies on digital forensic investigations.
Multi-modal detection covering images, video, and audio with detailed forensic breakdown.

Sensity’s platform is built for scenarios where a trained analyst examines a specific piece of media in detail, produces a forensic report, and that report is used in an investigation or prosecution.

The insurance problem:

Sensity’s forensic approach, while technically impressive, creates several challenges for insurance use cases:

35 minutes per manual review. Sensity’s forensic analysis workflow involves significant manual analyst time per item. For a law enforcement investigation examining a handful of critical pieces of evidence, this is acceptable. For an insurance carrier processing thousands of claims daily, it is not.
Insurance is an afterthought. Sensity’s published content includes approximately one generic blog post mentioning insurance. Their product development, sales materials, and case studies are oriented toward law enforcement, government, and media verification. Insurance is not a core market.
No claims workflow integration. Sensity’s platform operates independently of claims management systems. Results must be manually transferred into the claims workflow.
European focus. Sensity’s primary market and client base is European. Their understanding of Australian insurance regulations, APRA requirements, and local claims processes is limited.
No automated triage. Every item flagged for analysis requires manual forensic review. There is no automated tier that can process bulk claims media and filter down to items requiring human attention.
Pricing for forensic investigations. Sensity’s pricing reflects forensic-grade, analyst-intensive work. Per-analysis costs that make sense for criminal investigations become prohibitive when applied to every claims submission.

Deetech: Insurance-Scale Detection

Deetech approaches the deepfake detection problem from the opposite direction. Rather than starting with forensic investigation and trying to scale it for insurance, Deetech started with insurance operations requirements and built detection technology to meet them.

Designed for claims volume:

The fundamental difference is architectural. Deetech’s three-layer defense is built for the volume and speed of insurance claims:

Automated screening — Every media item attached to every claim is scanned automatically in seconds. No analyst involvement. No queue. This layer catches the obvious: known AI-generated signatures, recycled images from databases, metadata inconsistencies, and generation artifacts detectable by automated models.
Enhanced analysis — Items flagged by automated screening receive deeper multi-model analysis. Multiple detection approaches are applied in parallel, including frequency domain analysis, GAN fingerprint detection, diffusion model signatures, and environmental consistency checks. This layer runs in under a minute per item, still fully automated.
Forensic investigation — The small percentage of items that require human-level forensic analysis receive it. But because layers one and two have already filtered 95%+ of submissions, the forensic workload is manageable even at scale.

This architecture means an insurer processing 10,000 claims per day can screen every single media submission without creating a bottleneck. Sensity’s architecture would require thousands of analyst-hours per day for the same volume.

Production accuracy vs benchmark accuracy:

Sensity’s 98% accuracy on public datasets is a legitimate figure — on those datasets. But public deepfake benchmarks (FaceForensics++, DFDC, Celeb-DF) use high-quality, minimally compressed media that bears little resemblance to what appears in insurance claims.

Real claims media presents specific challenges:

Compression artifacts from WhatsApp, email, and web portals that obscure deepfake signatures
Low-resolution captures from older mobile devices common among some policyholder demographics
Metadata stripping by messaging platforms and email clients
Environmental noise — poor lighting, reflections, motion blur in dash cam footage
Document scans — photos of documents taken at angles with varying quality

Deetech’s models are trained and validated on media representative of actual claims submissions. Our production accuracy metrics reflect performance on compressed, noisy, real-world media — not laboratory conditions.

Detailed Comparison

Speed and Throughput

Metric	Sensity AI	Deetech
Per-item analysis time	~35 min (manual forensic)	Seconds (automated) to minutes (enhanced)
Daily capacity per analyst	~14 items	N/A — automated
Bulk claims processing	Not designed for bulk	Core capability
Catastrophe surge handling	Would require proportional analyst scaling	Automated scaling
Results delivery	Post-investigation report	Real-time, in claims workflow

For an insurer receiving 500 flagged media items per day, Sensity would require approximately 36 full-time analysts working continuously. Deetech processes the same volume automatically.

Accuracy Comparison

Both platforms are technically capable detectors. The relevant question is: accurate on what?

Sensity excels on high-quality, uncompressed media — the type encountered in law enforcement evidence and intelligence work.
Deetech is optimized for compressed, noisy, real-world claims media — the type submitted through claims portals, email, and messaging apps.

Neither platform’s accuracy claims are directly comparable because they measure performance on different media types. For insurance, the media types that matter are the ones policyholders actually submit.

Integration

Sensity operates as a standalone forensic platform. Analysis results live within Sensity’s interface and reporting system. Getting those results into a claims management system requires manual data entry or custom API integration work.

Deetech integrates natively with major claims platforms:

Guidewire ClaimCenter — detection results attached directly to claim records
Duck Creek Claims — automated flagging within existing workflow
Sapiens ClaimsPro — risk scores visible to adjusters in their standard interface
REST API — for carriers with custom or legacy claims systems

The practical impact: an adjuster using Deetech sees media authenticity information as part of their normal claims review. An adjuster using Sensity would need to leave their claims system, log into a separate platform, request analysis, wait for results, and manually record findings.

Forensic Reporting

This is where Sensity has genuine strength. Their forensic reports are detailed, technically rigorous, and designed for evidentiary use. For the small percentage of claims that proceed to formal investigation, litigation, or prosecution, this matters.

Deetech also produces forensic-grade reports, but designed specifically for insurance contexts:

Formatted for Australian regulatory standards (APRA, ASIC, AFCA)
Plain-language executive summaries alongside technical detail
Chain of custody documentation for digital evidence
Statistical confidence intervals calibrated for insurance decision-making
Exportable in formats compatible with common claims and investigation systems

For routine claims decisions, Deetech’s automated risk scores and summaries are sufficient. For contested claims, Deetech’s full forensic reports meet the same evidentiary bar — with the added benefit of insurance-specific formatting and regulatory alignment.

Threat Intelligence

Sensity maintains an impressive threat intelligence operation, tracking new deepfake techniques, monitoring dark web marketplaces, and publishing research on emerging generative AI capabilities. This intelligence is valuable for understanding the threat landscape.

Deetech’s threat intelligence is focused specifically on insurance fraud trends:

New techniques being used in claims fraud
Regional patterns in AI-generated claims media
Emerging tools accessible to opportunistic fraudsters (not just sophisticated actors)
Catastrophe event fraud patterns specific to the insurance sector

For a CISO or security team, Sensity’s broad threat intelligence is relevant. For a claims operations leader, Deetech’s insurance-focused intelligence is more actionable.

The Scalability Question

Insurance fraud detection must operate at portfolio scale. A typical mid-size Australian insurer processes hundreds of thousands of claims annually. A global carrier processes millions.

Sensity’s forensic model works when you’re investigating individual cases — a suspected fraud ring, a high-value contested claim, a specific piece of evidence for prosecution. It does not work when you need to screen every claim submission for AI-generated content.

This isn’t a criticism of Sensity’s technology. It’s a recognition that their product was designed for a different use case. A forensic laboratory and a production screening facility serve different purposes, even if the underlying science overlaps.

Deetech is the production screening facility. Every claim, every media item, every time. The deepfake detection FAQ covers how this automated approach handles common concerns about false positives and processing speed.

Cost Implications

Sensity’s forensic analysis pricing reflects the analyst time and expertise involved. Per-item costs appropriate for criminal investigations become economically challenging when applied to insurance volumes.

Consider a carrier processing 200,000 claims annually with an average of 3 media items per claim. That’s 600,000 items requiring screening. At forensic investigation pricing — even at heavily discounted rates — the annual cost would exceed most carriers’ entire fraud detection budgets.

Deetech’s per-claim pricing is designed for insurance economics. Automated screening of all submissions at a fraction of the cost of manual forensic review, with full forensic analysis reserved for the small percentage of items that warrant it.

When Sensity AI Makes Sense

Sensity is a strong choice for:

Law enforcement agencies investigating specific deepfake-related crimes
Legal teams needing forensic evidence for active litigation
Intelligence organizations tracking deepfake threats at a strategic level
Media organizations verifying specific pieces of contested content
One-off investigations where depth of analysis matters more than speed

When Deetech Is the Better Choice

Deetech is the right solution for insurance carriers that need:

Automated screening of all claims media at scale
Detection integrated into existing claims workflows
Insurance-specific detection models and accuracy metrics
Cost-effective per-claim pricing aligned with insurance economics
Forensic reports formatted for insurance regulatory requirements
Catastrophe event fraud detection capabilities

The Verdict

Sensity AI and Deetech are both technically capable deepfake detection platforms. They are not, however, substitutes for each other.

Sensity is a forensic investigation tool. Deetech is an insurance operations platform. Choosing between them depends entirely on whether you need to investigate individual cases in depth or screen an entire claims portfolio automatically.

For insurance carriers, the answer is almost always the latter. And for that use case, Deetech was purpose-built from day one.

For a broader view of how these platforms compare with other options, see our top deepfake detection tools for insurance in 2026.

To learn how deetech helps insurers detect deepfake fraud with purpose-built AI detection, visit our solutions page or request a demo.