Deepfakes Are Coming for Your Bank Account

8 hours ago

OpenAI's ChatGPT Images 2.0 can create photorealistic deepfakes with convincing text, enabling fraudulent images like fake TikTok screenshots, prescriptions, bank alerts, receipts, and IDs.
The model excels at generating fake screenshots and documents, which could amplify phishing scams by tricking targets into clicking malicious links or revealing sensitive information.
Despite prohibitions on misuse, both OpenAI and Google's image-generation tools have guardrails that are easily bypassed, with metadata or watermarks offering limited protection against fraud.
AI-generated deepfakes extend beyond political scandals to mundane scams targeting individuals, posing greater risks than fake news due to their micro-targeted and persuasive nature.
Experts warn that fraud prevention lags behind advancing AI capabilities, with open-source models and technological fixes struggling to keep pace with scammers' creativity.

Hasty Briefsbeta