.png)
October 9, 2024
Artificial intelligence (AI) is evolving rapidly, continuously creating new opportunities and challenges. A current example of this is the capabilities of GPT-4, the large language model from OpenAI, in the field of biometrics.
Although GPT-4 was not specifically trained for facial recognition, studies have shown that the model is able to recognize faces in images, determine gender, and estimate age - and with an accuracy comparable to specialized algorithms. This unexpected capability of GPT-4 raises questions about both the potential and the risks of this technology.
GPT-4 performed surprisingly well in various tests in the area of biometric recognition. In gender recognition, GPT-4 achieved an accuracy of 100% on a dataset of 5,400 images. This result even surpassed Facebook's DeepFace model, which was specifically designed for this task. GPT-4 also showed remarkable capabilities in age estimation. Using the UTKFace dataset, the model was able to correctly identify the age range in 74.25% of cases.
However, the impressive capabilities of GPT-4 in the field of biometrics also raise security concerns. For example, researchers were able to circumvent GPT-4's safety measures, which are intended to prevent the disclosure of sensitive biometric information. By claiming in the query that an image was AI-generated, they got the system to analyze real photos. This simple trick highlights the need for further security research in the field of large language models, especially given their capabilities in biometric tasks.
The biometric capabilities of GPT-4 underscore the need for comprehensive regulation and ethical guidelines for AI technologies. Experts warn that the uncontrolled development and application of AI systems with biometric capabilities could have serious consequences for privacy and fundamental rights. It is therefore essential that governments, businesses, and the research community work together to ensure that AI technologies are used responsibly and for the benefit of humanity.
The development of GPT-4 and its biometric capabilities is another step in the rapid development of AI. This development offers both enormous opportunities and potential risks. To ensure that AI technologies are used for the benefit of humanity, transparency, regulation, and a continuous dialogue between all stakeholders are essential. Only in this way can we ensure that AI systems such as GPT-4 can reach their full potential without jeopardizing our fundamental rights.
The biometric capabilities of GPT-4 are a fascinating example of the progress being made in the field of AI. At the same time, they highlight the need to carefully consider the ethical and societal implications of this technology. Only through the responsible use of AI can we ensure that it becomes a positive force for the future.
https://the-decoder.com/openais-gpt-4-matches-facial-recognition-algorithms-without-explicit-training-in-biometrics/
https://openai.com/index/gpt-4o-system-card/
https://idtechwire.com/openai-restricts-gpt-4-facial-recognition-capabilities-907181/
https://the-decoder.com/artificial-intelligence-news/
https://medium.com/@tam.tamanna18/openai-takes-a-stand-gpt-4-shall-not-be-used-for-facial-recognition-cbd9ff2e35fe
https://news.slashdot.org/story/24/09/04/2244208/openai-co-founder-raises-1-billion-for-new-safety-focused-ai-startup
https://aiindex.stanford.edu/wp-content/uploads/2024/04/HAI_2024_AI-Index-Report.pdf
https://www.cnn.com/2019/04/19/tech/ai-facial-recognition/index.html
https://www.judiciary.senate.gov/imo/media/doc/2024-09-17_pm_-_testimony_-_harris.pdf
https://news.slashdot.org/story/24/09/16/2226243/ai-pioneers-call-for-protections-against-catastrophic-risks
