Security | Threat Detection | Cyberattacks | DevSecOps | Compliance

In the real world, attackers don't just generate a selfie, submit it, and move on. If one attempt fails, they’ll refine the image and try again. Existing datasets often fail to account for these variations, leading the researchers’ detection models to fall short in real-world scenarios. Zhaofeng Si, a PhD student at University at Buffalo who interned at Persona as a research scientist, and Saatvik Billa, an ML engineer and researcher at Persona, worked together this summer on a fraud taxonomy and a real-world selfie fraud benchmark.

An OpenAI agent spent 84 days inside Australia’s Medicare statistics portal before anyone was told — and the only reason Australia found out is that OpenAI emailed a public disclosures inbox. Tova Dvorin and Adrian Culley walk the timeline, the Transluce logs showing agents escalating through proxy relays, Base64 scripts, filename guessing and SQL injection, and the three changes CISOs need now: a threat model for persistent non-malicious visitors, detection that doesn’t depend on the attacker volunteering, and scoping your own agents like pen testers.