Navigating the reality of AI images: tips for identifying deepfakes

Navigating the Reality of AI Images: Tips for Identifying Deepfakes

Artificial intelligence (AI) has firmly established itself in the domain of visual media, introducing the capacity to create hyper-realistic and sometimes deceptive images. The surge in AI technology has made it challenging to discern authentic photographs from manipulated ones. As this technology evolves, the importance of recognising AI-generated visuals, often referred to as ‘deepfakes’, has become paramount.

AI Image Generation: An Overview

AI editing tools have revolutionised how images are created and manipulated. These tools can manufacture images that appear uncannily real, which can be particularly problematic when they distort someone’s likeness. For instance, Google’s Add Me feature on the Pixel 9 phone allows photographers to insert themselves into images, creating a lifelike yet potentially misleading result.

Platforms such as Midjourney simplify AI image generation, enabling the production of highly realistic images. A notable example is last year’s viral image of the Pope in a white Balenciaga puffer jacket, which successfully deceived many viewers.

Key Strategies to Identify Deepfakes

To navigate the influx of AI-driven visuals, here are some essential strategies to help in discerning genuine images from deepfakes:

1. Zoom in on Details:

   • Examine small details within an image, as inconsistencies often become apparent upon closer inspection.
   • AI-generated hands and fingers frequently contain noticeable errors. Additionally, faces in deepfakes may display blurriness around the edges.
   • In videos, check for lip-sync issues that may suggest manipulation.

2. Assess Emotional Expression:

   • AI tools tend to struggle with the nuances of complex facial expressions. A smile that appears too rigid or a face that looks unusually neutral can be indicative of AI intervention.
   • Pay attention to expressions in videos as well, as mismatches between facial expressions and spoken words can be revealing.

3. Evaluate the Overall Image:

   • AI images often appear too perfect, lacking the natural imperfections found in real photographs.
   • Notice lighting and shadow discrepancies that seem too uniform across different subjects in a group photo, or a plasticky aesthetic that feels artificial.

4. Scrutinise the Background:

   • Examine backgrounds for any inconsistencies or architectural elements that defy logic.
   • AI can generate backgrounds that have physical contradictions or unrealistic features.

5. Check the Real-World Context:

   • Conduct a web search to verify if the image aligns with the real-world context, especially for public figures.
   • Determine if the purported events depicted in the image were reported by credible sources around the time the image was supposedly taken.
   • For social media images of less notable individuals, employ a critical eye and think carefully before drawing conclusions.

Conclusion

As AI technology continues to advance, the capabilities to generate and manipulate images will only become more sophisticated. Staying informed and vigilant remains crucial for users to navigate this digital landscape effectively. By applying these strategies, individuals can better distinguish between genuine images and deepfakes, thereby mitigating the risk of being misled by convincingly fabricated visuals.

Source: Noah Wire Services