work
What I look at
- Deceptive media. Ads, endorsements and claims built to leave a false impression: celebrity deepfakes, invented customers, "as seen in" strips, guarantees the fine print takes back.
- Coordinated inauthentic behaviour. Accounts pretending to be someone they aren't, usually several at once: persona profiles, comment rings under ads, review farms, blue checks bought to look real.
- The infrastructure underneath. Domains, certificates, hosting, ad accounts and payment rails. A persona can lie about itself. Its records can't.
How I see it
I work from public surfaces: registries, DNS and certificate records, archived pages, ad libraries and public posts.
Where it helps, I engage first-hand. Products marketed to Canadians get bought and run on honeypots, so I can watch what they actually do. Scam operations get drawn into long conversations with bots, which wastes their time and shows how they run.
AI red teaming
The same habit, pointed at AI products: I test what an LLM feature will actually do when someone asks it the question it was told never to answer. Case in point: Kali GPT. Hire me for it through Upwork.
Where it goes
Some of the research becomes public. Most of it goes privately to the people who can act on it. See outputs.