Slop or Not, a new AI text and image detection application for iPhone and Mac, operates entirely offline and does not require an account. The app leverages on-device AI models, powered by the Apple Neural Engine, to identify AI-generated content.
The development of such tools comes as a recent survey indicates a growing struggle among individuals to differentiate authentic content from AI-generated material. The survey found that 85% of respondents reported difficulty in this area, an increase from 66% just a year prior. Half of those surveyed also reported encountering AI-driven scams, including deepfakes, voice cloning, fake reviews, and personalized phishing attacks, leading many to reduce their online sharing. This trend suggests that increasingly realistic AI content is making scams harder to identify and is eroding trust online.
Slop or Not provides a probability score indicating the likelihood of an image being AI-generated. A key feature is its ability to detect Google's invisible SynthID watermark, allowing compatible images to be verified directly on the device. When a SynthID watermark is identified, it offers additional evidence that an image originated from AI. During testing, scanning a credit card photo resulted in the app reporting it as "Likely Watermarked."
In practical tests, the app successfully identified a highly stylized AI-generated image of a backyard swimming pool with over 93% confidence. Another AI-generated image of a garden with a wooden bench, which was described as more convincing, resulted in a near-even split: 50.4% AI-generated versus 49.6% real. Despite the close percentages, the app still classified it as "Probably AI Slop," indicating that some images fall into a gray area rather than yielding a definitive result. Conversely, two genuine photographs—one of tall trees in a forest and a black-and-white concert photo—were correctly identified as real.
The application also includes a quiz designed to test a user's ability to distinguish AI-generated images from real ones. In one instance, an average user correctly identified six out of ten images, while four images successfully fooled them.
Slop or Not is designed to analyze various forms of content, including suspicious social media posts, AI-generated portraits, manipulated photos, product images, and screenshots shared via messaging apps. Its offline operation is particularly beneficial for users concerned about privacy, as it avoids uploading sensitive files to cloud-based services. Potential users include students and educators reviewing essays for AI-generated text, journalists and fact-checkers verifying images, and businesses checking reports, marketing copy, or confidential documents.
While tools like Slop or Not offer an additional layer of protection against scams and misinformation, it is acknowledged that AI detection is not an exact science and no tool can guarantee perfect accuracy. Experts recommend a multi-faceted approach to reduce risk, including verifying unexpected requests through trusted channels, using AI detection tools, enabling multi-factor authentication, limiting personal media shared online, and staying informed about new deepfake techniques. The app is presented as a complement to critical thinking and human judgment, rather than a replacement.
