HARMFUL
BEHAVIOUR DETECTION IN LANGUAGE

AI agents that surface grooming, coercion and victim-blaming across your conversations, so people can stay safer, organisations reduce losses, and regulators can trust your decisions.

VICTIM BLAMING

Victim blaming in official and organisational communications can echo the harm people have already experienced and erode trust in your service.

Our victim-blaming detection app ViDA, plus custom APIs, help you evidence victim-blaming language at scale and track how policies, culture and training programmes are changing it over time.

Learn more about how it works.

TRY IT OUT

BOOK A CALL

What We Build

We’re building AI agents that understand harmful behaviour in language, from victim-blaming and coercive control to APP and romance fraud, not just keywords. Our platform links information across messages, notes and reports (chats, emails, call notes, case files) to surface patterns of coercion so humans can make safer, more empathetic and financially sound decisions.

Agentic AI APIs

Our AI agents read conversations the way a trained specialist would: looking for patterns over time, shifts in tone, and hidden power dynamics. Each agent focuses on a specific type of harmful behaviour, and they work together to give you a clear, contextual picture of risk.

Linked information, not isolated messages

Harmful behaviour rarely shows up in a single sentence. Our system links information across chats, emails, call notes and case files to detect grooming, coercion and victim-blaming as they build up. This linking makes our detectors more accurate, and helps you act sooner with more confidence.

Built for trust and safety

Every flag our agents raise comes with an explanation you can inspect, challenge and improve. We combine behavioural psychology with transparent AI methods, so your teams can trust what they see, your users stay safer, and regulators can understand how decisions were made.

Choose Empathy

Let our trauma-informed AI handle the heavy lifting of identifying harmful patterns in language, so your experts can focus on listening, supporting and protecting.

Who We Help

herEthical AI works with organisations who need to understand harmful behaviour in language, not just numbers.

Banks, PSPs and fintechs

Detect APP and romance fraud by analysing the language in customer communications before and after payments. Help customers avoid life-changing losses, reduce avoidable reimbursement costs, and support quicker, fairer decisions with explainable evidence that stands up to regulatory scrutiny.

Justice, policing and regulators

Surface harmful, non-client-focused language and patterns of coercion in evidence, case files, clinical notes, support records, or advice channels. Helping staff understand complex cases and adopt a trauma-informed approach.

Support, health and advice services

Surface harmful, non-client-focused language and patterns of coercion in clinical notes, support records or advice channels, without labelling staff as “perpetrators”.

Clients

WhatsApp Image 2025-06-20 at 12.44_edited.jpg

What you can do with herEthical AI

Our multi-agent platform turns subtle patterns in language into clear signals your team can act on, cutting investigation time and preventing avoidable losses for you and your customers.

Detect APP & romance fraud earlier

Analyse chat logs, call notes and CRM entries for grooming and coercion patterns, so you can intervene before money leaves the account or is irretrievably lost. Prioritise risky cases, reduce unnecessary reimbursements, support five-day reimbursement timelines and produce defensible evidence files.
Independent and private checks
You can also offer customer-facing tools that let people quietly check risky conversations themselves before sending money.

Audit victim-blaming and bias at scale
Run detectors across reports, statements or judgments to understand where harmful language appears, how it changes over time and where training or policy interventions are working.
Support people experiencing coercive control

Help survivors organise their communications into timelines, highlight coercive tactics and create annotated bundles they can use with support workers, lawyers or, if they choose, the police.
Custom linguistic harm detectors

Bring us your hardest language problem – from workplace harassment to predatory sales scripts – and we’ll help design and build detectors from our library of behavioural components.